Papers with Responsible & Trustworthy NLP
Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 2: Short Papers) (2026.eacl-short)
Copied to clipboard
| Challenge: | EACL 2026 is the first time the conference is being hosted in the African continent . organisers of the event have been able to accommodate a very compressed timeframe . |
| Approach: | a message from the General Chair of the European Chapter of the Association for Computational Linguistics . a grateful to the many volunteers whose dedication, generosity have made this conference possible . |
| Outcome: | the 19th Conference of the European Chapter of the Association for Computational Linguistics (EACL 2026) will take place in Rabat, Morocco, in March 24-29, 2026. |
Revisiting Zero-Shot Abstractive Summarization in the Era of Large Language Models from the Perspective of Position Bias (2024.naacl-short)
Copied to clipboard
| Challenge: | Position bias is a tendency of a model unfairly prioritizing information from certain parts of the input text over others, leading to undesirable behavior. |
| Approach: | They propose to measure position bias in large language models for zero-shot summarization tasks by measuring position bias. |
| Outcome: | The proposed model performance and position biases lead to new insights and discussion on zero-shot summarization tasks. |
How Does Stereotype Content Differ across Data Sources? (2024.starsem-1)
Copied to clipboard
| Challenge: | Existing studies of stereotypes using rating scales capture beliefs and opinions about different social groups. |
| Approach: | They compare stereotype-relevant measures of social group social status with traditional scales and a word-list generation task using free-text data. |
| Outcome: | The results compare with traditional surveys and a spontaneous word-list generation task. |
Countering Hateful and Offensive Speech Online - Open Challenges (2024.emnlp-tutorials)
Copied to clipboard
Leon Derczynski, Marco Guerini, Debora Nozza, Flor Miriam Plaza-del-Arco, Jeffrey Sorensen, Marcos Zampieri
| Challenge: | a comprehensive understanding of the field is needed to maintain respectful and inclusive online environments. |
| Approach: | This tutorial aims to provide attendees with a comprehensive understanding of the field by delving into essential dimensions such as multilingualism, counter-narrative generation, a hands-on session with one of the most popular APIs for detecting hate speech, fairness, and ethics in AI, and the use of recent advanced approaches. |
| Outcome: | This tutorial aims to provide attendees with a comprehensive understanding of the field by delving into essential dimensions such as multilingualism, counter-narrative generation, a hands-on session with one of the most popular APIs for detecting hate speech, fairness, and ethics in AI, and the use of recent advanced approaches. |
Fact Checking or Psycholinguistics: How to Distinguish Fake and True Claims? (D19-66)
Copied to clipboard
| Challenge: | Using psycholinguistic features to distinguish lies from true statements is a difficult task and a problem to be solved. |
| Approach: | They compare psycholinguistic text features with fact checking approaches to distinguish lies from true statements using data from a large ongoing study. |
| Outcome: | The proposed methods outperform both fact checking and human baselines but the accuracy is not high. |
Attention is not not Explanation (D19-1)
Copied to clipboard
| Challenge: | Attention mechanisms play a central role in NLP systems, especially within recurrent neural network (RNN) models. |
| Approach: | They propose to use a simple uniform-weights baseline, a variance calibration and a diagnostic framework to determine when/whether attention can be used as explanation in RNN models. |
| Outcome: | The proposed tests show that even reliable adversarial distributions don't perform well on the simple diagnostic, indicating that prior work does not disprove the usefulness of attention mechanisms for explainability. |
Resisting the Lure of the Skyline: Grounding Practices in Active Learning for Morphological Inflection (2024.acl-short)
Copied to clipboard
| Challenge: | Several approaches to active learning are available, including confidence-based, diversity-based and committee-based. |
| Approach: | They propose to use a baseline and a skyline to measure the accuracy of the unannotated sample pool. |
| Outcome: | The proposed model outperforms a random selection baseline and a skyline approach. |
FairLib: A Unified Framework for Assessing and Improving Fairness (2022.emnlp-demos)
Copied to clipboard
| Challenge: | Existing approaches to assess and improve model fairness have been inconsistent and inconsistent. |
| Approach: | They propose an open-source python library for assessing and improving model fairness. |
| Outcome: | The proposed framework can be used for natural language, images, and audio. |
Entity-Aware Dual Co-Attention Network for Fake News Detection (2023.findings-eacl)
Copied to clipboard
| Challenge: | Existing models for fake news detection are limited in their ability to detect it from different aspects. |
| Approach: | They propose a Dual Co-Attention Network (Dual-CAN) for fake news detection that takes news content, social media replies, and external knowledge into consideration. |
| Outcome: | The proposed model outperforms existing models in two benchmark datasets. |
Unsupervised Cross-Lingual Representation Learning (P19-4)
Copied to clipboard
| Challenge: | a comprehensive survey of cutting-edge weakly-supervised and unsupervised cross-lingual word representations is presented . |
| Approach: | This tutorial provides a comprehensive survey of recent work on weakly-supervised and unsupervised cross-lingual word representations. |
| Outcome: | This tutorial provides a comprehensive survey of cutting-edge weakly-supervised and unsupervised word representations. |
Clinical Text Anonymization, its Influence on Downstream NLP Tasks and the Risk of Re-Identification (2023.eacl-srw)
Copied to clipboard
| Challenge: | De-identification and anonymization of clinical data is needed to solve access to clinicaldata. |
| Approach: | They propose to use text anonymization techniques to break the anonymization of clinical data . they propose to apply a re-identification attack to the anonymized text data to break this. |
| Outcome: | The proposed approach can break the anonymization of clinical data, the authors show . |
TALEN: Tool for Annotation of Low-resource ENtities (P18-4)
Copied to clipboard
| Challenge: | Named entity recognition (NER) is a task that requires a large amount of training data and annotators who do not speak the language are hard or impossible to find. |
| Approach: | They propose a web-based interface for named entity annotation in low-resource settings . TALEN includes in-place lexicon integration, TF-IDF token statistics, Internet search, and entity propagation . |
| Outcome: | The proposed interface performs better than a popular annotation tool and is more accurate and recall-rich than the current one. |
ReasonGraph: Visualization of Reasoning Methods and Extended Inference Paths (2025.acl-demo)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) reasoning processes are complex and lack of organized visualization tools creates barriers to understanding, evaluation, and improvement. |
| Approach: | They propose a web-based platform for visualizing and analyzing LLM reasoning processes. |
| Outcome: | The proposed platform shows high parsing reliability, efficient processing, and excellent usability across various downstream applications. |
Dis2Dis: Explaining Ambiguity in Fact-Checking (2025.findings-naacl)
Copied to clipboard
| Challenge: | Ambiguity is a linguistic tool for encoding information efficiently, yet it also causes misunderstandings and disagreements. |
| Approach: | They propose a constrained generation task for explaining ambiguous claims in fact-checking by editing them to spell out an interpretation that can be unequivocally supported by the given evidence. |
| Outcome: | The proposed model disambiguates claims 72% of the time compared to a simple copy baseline and a Large Language Model baseline. |
Learning to Verify Summary Facts with Fine-Grained LLM Feedback (2025.coling-main)
Copied to clipboard
| Challenge: | Recent advances in large language models (LLMs) have significantly enhanced the text summarization performance, but hallucination issues still occur in summaries. |
| Approach: | They propose a large-scale dataset containing fine-grained factual feedback on summaries that can be fine tuned by using Large Language Models (LLMs) they employ 10 distinct LLMs for diverse summary generation and Llama-3-70B-Instruct for feedback. |
| Outcome: | The proposed model outperforms models trained on smaller human-annotated datasets while maintaining high performance. |
Learning Domain Terms - Empirical Methods to Enhance Enterprise Text Analytics Performance (2020.coling-industry)
Copied to clipboard
| Challenge: | Standard text analytics algorithms are degraded on consumer generated data, a study finds . enterprise data contains domain terms that are not available in general purpose thesaurus . |
| Approach: | They propose unsupervised methods for cleaning noisy repositories after domain terms are segregated from true noise terms. |
| Outcome: | The proposed method improves classification accuracy after noise correction . it is based on a systematic analysis of enterprise data . |
Learning Interpretable Latent Dialogue Actions With Less Supervision (2022.aacl-main)
Copied to clipboard
| Challenge: | supervised neural dialogue modeling requires a significant amount of work to obtain turn-level labels, usually with dialogue state annotation. |
| Approach: | They propose a novel architecture for explainable modeling of task-oriented dialogues with discrete latent variables to represent dialogue actions. |
| Outcome: | The proposed model outperforms previous approaches with less supervision in terms of perplexity and BLEU on three datasets. |
Improving Cross-Domain Low-Resource Text Generation through LLM Post-Editing: A Programmer-Interpreter Approach (2024.findings-eacl)
Copied to clipboard
| Challenge: | Large pre-trained language models such as GPT-3.5 and GPT-4 have gained significant attention in natural language research due to limited computational resources or inaccessible parameters. |
| Approach: | They propose a neural programmer-interpreter approach that preserves the domain generalization ability of LLMs while editing their output. |
| Outcome: | The proposed framework significantly improves GPT-3.5’s performance in logical form-to-text conversion and low-resource machine translation, surpassing other state-of-the-art (SOTA) LLM post-editing methods in cross-domain settings. |
Small Models, Big Impact: Efficient Corpus and Graph-Based Adaptation of Small Multilingual Language Models for Low-Resource Languages (2025.acl-srw)
Copied to clipboard
| Challenge: | Low-resource languages (LRLs) face significant challenges in natural language processing due to limited data. |
| Approach: | They evaluate adapter-based methods for adapting mLMs to low-resource languages . they use unstructured text and structured knowledge from ConceptNet to evaluate adapters . |
| Outcome: | The proposed methods outperform large language models and LLaMA-3 and deepSeek-R1 models on low training data. |
SAFARI: A Community-Engaged Approach and Dataset of Stereotype Resources in the Sub-Saharan African Context (2026.eacl-short)
Copied to clipboard
Aishwarya Verma, Laud Ammah, Olivia Nercy Ndlovu Lucas, Andrew Zaldivar, Vinodkumar Prabhakaran, Sunipa Dev
| Challenge: | Existing data collection approaches to generative AI are inadequate to assess its safety and utility. |
| Approach: | They propose a multilingual stereotype resource that uses socioculturally-situated, community-engaged methods to assess the region’s linguistic diversity and traditional orality. |
| Outcome: | The proposed method covers four sub-Saharan African countries that are severely underrepresented in NLP resources: Ghana, Kenya, Nigeria, and South Africa. |
All-in-One: A Deep Attentive Multi-task Learning Framework for Humour, Sarcasm, Offensive, Motivation, and Sentiment on Memes (2020.aacl-main)
Copied to clipboard
| Challenge: | Empirical results show the efficacy of our proposed multi-task framework over existing state-of-the-art systems. |
| Approach: | They propose a multi-task, multi-modal deep learning framework to solve multiple tasks simultaneously. |
| Outcome: | The proposed framework performs better than existing state-of-the-art systems on a complicated form of information, i.e., memes. |
Personalized Microblog Sentiment Classification via Adversarial Cross-lingual Multi-task Learning (D18-1)
Copied to clipboard
| Challenge: | Existing personalized microblog sentiment classification methods suffer from the insufficiency of discriminative tweets for personalization learning. |
| Approach: | They propose to use user-attention-based Convolutional Neural Networks to capture individuality and opinion bias in microblog posts and a novel adversarial cross-lingual learning framework to enrich the user post representation. |
| Outcome: | The proposed method outperforms state-of-the-art baseline algorithms with large margins on English and Chinese microblog datasets. |
Parameter-free and Accessible Prompt Learning to Enhance Adversarial Robustness for Pre-trained Vision-Language Models (2025.naacl-long)
Copied to clipboard
| Challenge: | Large pre-trained Vision-Language Models (VLMs) have revolutionized downstream vision-language tasks including classification, object detection, and segmentation. |
| Approach: | They propose to search for text prompts at the word level rather than optimizing continuous textual embeddings to boost adversarial robustness. |
| Outcome: | Experiments show that the proposed method outperforms hand-engineered prompts with average gains of +4.9% and +5.8%. |
A Novel Workflow for Accurately and Efficiently Crowdsourcing Predicate Senses and Argument Labels (2020.findings-emnlp)
Copied to clipboard
| Challenge: | Prior attempts to develop crowdsourcing methods have either had low accuracy or required substantial expert annotation. |
| Approach: | They propose a multi-stage crowd workflow that reduces expert involvement without sacrificing accuracy. |
| Outcome: | The proposed method reduces expert effort by 4x, from 56% to 14% of cases. |
Self-Supervised Meta-Learning for Few-Shot Natural Language Classification Tasks (2020.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for supervised meta-learning require many training tasks to generalize . cloze-style objectives can be used to generate a large, rich, meta-training task distribution from unlabeled text. |
| Approach: | They propose a self-supervised approach to generate a large, rich, meta-learning task distribution from unlabeled text. |
| Outcome: | The proposed approach generates a large, rich, meta-learning task distribution from unlabeled text. |
Multitask Semi-Supervised Learning for Class-Imbalanced Discourse Classification (2021.emnlp-main)
Copied to clipboard
| Challenge: | Discourse learning is a complex task, and schemas evolve across annotation efforts preventing compilation of smaller datasets into larger ones. |
| Approach: | They propose a multitask learning approach that can combine discourse datasets from similar and diverse domains to improve discourse classification. |
| Outcome: | The proposed approach improves on the NewsDiscourse dataset by 4.9% over current state-of-the-art benchmarks on one of the largest discourse datasets. |
STAR: Spectral Truncation and Rescale for Model Merging (2025.naacl-short)
Copied to clipboard
| Challenge: | Model merging is an efficient way of obtaining a multi-task model from several pretrained models without further fine-tuning. |
| Approach: | They propose a model merging technique that aims at mitigating "merging conflicts" by truncating small components in the respective spectral spaces and then an automatic parameter rescaling scheme to retain the nuclear norm of the original matrix. |
| Outcome: | The proposed model outperforms baseline models on flan-T5 by 4.2% and is robust to hyperparamater choice. |
Scaling Neural ITN for Numbers and Temporal Expressions in Tamil: Findings for an Agglutinative Low-resource Language (2023.emnlp-industry)
Copied to clipboard
| Challenge: | Inverse Text Normalisation (ITN) is a textrewriting task that converts verbalized text to written form. |
| Approach: | They propose to use a seq2seq model, a non-autoregressive text editor and a sequence tagger + rules combination to fine-tune three pre-trained neural models. |
| Outcome: | The proposed model improves with bootstrapping and data augmentation, and bootstrap alone shows a percentage improvement of 14.12 %. |
Fact Checking with Insufficient Evidence (2022.tacl-1)
Copied to clipboard
| Challenge: | Existing work on how to automate fact checking relies on information obtained from external sources. |
| Approach: | They propose a fluency-preserving method for omitting information from the evidence at the constituent and sentence level and a diagnostic dataset for FC with omitted evidence. |
| Outcome: | The proposed method improves evidence sufficiency prediction by 17.8 F1 score and 2.6 F1 scores. |
SPARTA: Efficient Open-Domain Question Answering via Sparse Transformer Matching Retrieval (2021.naacl-main)
Copied to clipboard
| Challenge: | SPARTA is a novel neural retrieval method for open-domain question answering . it learns a sparse representation that can be efficiently implemented as an Inverted Index . |
| Approach: | They propose a method that learns a sparse representation that can be implemented as an Inverted Index. |
| Outcome: | The proposed method achieves state-of-the-art results on 4 open-domain question answering tasks and 11 retrieval question answering (ReQA) tasks. |
Parameter-efficient Multi-task Fine-tuning for Transformers via Shared Hypernetworks (2021.acl-long)
Copied to clipboard
| Challenge: | State-of-the-art parameter-efficient fine-tuning methods rely on introducing adapter modules between the layers of a pretrained language model. |
| Approach: | They propose a framework that can learn adapter parameters for all layers and tasks by generating them using shared hypernetworks. |
| Outcome: | The proposed framework improves performance on the well-known GLUE benchmark while adding only 0.29% parameters per task. |
Implicitly Abusive Language – What does it actually look like and why are we not getting there? (2021.naacl-main)
Copied to clipboard
| Challenge: | Existing datasets make learning implicit abuse difficult, argues a new position paper . a lack of work on implicit abuse has limited the effectiveness of automatic detection . |
| Approach: | They argue that existing datasets make learning implicit abuse difficult . they propose a divide-and-conquer strategy to detect implicit abuse . |
| Outcome: | The proposed model could be improved to detect implicit abuse in a dataset with a standardized model. |
Shoulders of Giants: A Look at the Degree and Utility of Openness in NLP Research (2024.acl-short)
Copied to clipboard
| Challenge: | We analysed a sample of NLP research papers published in ACL Anthology . we observe a wide language-wise disparity in publicly available NLP-related artefacts . |
| Approach: | They analysed NLP research papers archived in ACL Anthology to quantify degree of openness and benefit of such an open culture in the NLP community. |
| Outcome: | The results show that more than 30% of the papers published in ACL Anthology do not release their artefacts publicly. |
Unsupervised Learning of Discourse-Aware Text Representation for Essay Scoring (P19-2)
Copied to clipboard
| Challenge: | Existing document embedding approaches focus on capturing sequences of words in documents . however, some document classification and regression tasks need to consider discourse structure of text . |
| Approach: | They propose an unsupervised approach to capture discourse structure in terms of coherence and cohesion for document embedding that does not require expensive parsers or annotation. |
| Outcome: | The proposed method improves essay Organization scoring and Argument Strength scoring. |
Subjective Crowd Disagreements for Subjective Data: Uncovering Meaningful CrowdOpinion with Population-level Learning (2023.acl-long)
Copied to clipboard
| Challenge: | Annotator disagreements are resolved before learning takes place, but researchers question the performance of a system when annotators disagree. |
| Approach: | They propose a method that uses language features and label distributions to pool similar items into larger labels. |
| Outcome: | The proposed method is based on five publicly available datasets with varying levels of disagreements on social media and in the wild using a dataset from Facebook. |
Unsupervised Abstractive Opinion Summarization by Generating Sentences with Tree-Structured Topic Guidance (2021.tacl-1)
Copied to clipboard
| Challenge: | Abstractive summarization is a novel method for opinionated texts . it uses a recursive Gaussian mixture to generate topic sentences . |
| Approach: | They propose an unsupervised abstractive summarization method for opinionated texts . they alternate the unimodal Gaussian prior with a recursive Gausssian mixture . |
| Outcome: | The proposed method generates topic sentences with tree-structured topic guidance, which are more informative and cover more input contents than the current model. |
ProoFVer: Natural Logic Theorem Proving for Fact Verification (2022.tacl-1)
Copied to clipboard
| Challenge: | Recent fact verification systems rely on neural network classifiers for veracity prediction, which lack explainability. |
| Approach: | They propose a model that generates natural logic-based inferences as proofs using lexical mutations between spans in the claim and the evidence retrieved. |
| Outcome: | The proposed model has highest label accuracy and second best score in the FEVER leaderboard. |
A Prompt Array Keeps the Bias Away: Debiasing Vision-Language Models with Adversarial Learning (2022.aacl-main)
Copied to clipboard
| Challenge: | Large-scale, pretrained vision-language models are growing in popularity due to impressive performance on downstream tasks with minimal finetuning. |
| Approach: | They propose to apply ranking metrics to image-text representations to investigate bias measures and debiasing methods to reduce various bias measures. |
| Outcome: | The proposed model reduces bias measures with minimal degradation to image-text representations. |
PromptPrism: A Linguistically-Inspired Taxonomy for Prompts (2026.findings-eacl)
Copied to clipboard
| Challenge: | PromptPrism is a linguistically-inspired taxonomy that enables prompt analysis across three hierarchical levels. |
| Approach: | They propose a linguistically-inspired taxonomy that enables prompt analysis across three hierarchical levels: functional structure, semantic component, and syntactic pattern. |
| Outcome: | The proposed taxonomy bridges traditional language understanding with modern LLM research . it improves prompt quality and improves model performance across tasks . |
Multilingual Speech Translation from Efficient Finetuning of Pretrained Models (2021.acl-long)
Copied to clipboard
Xian Li, Changhan Wang, Yun Tang, Chau Tran, Yuqing Tang, Juan Pino, Alexei Baevski, Alexis Conneau, Michael Auli
| Challenge: | Recent advances in text pretraining and finetuning have improved multitasking applications significantly. |
| Approach: | They propose a minimalistic LNA finetuning approach to build multilingual speech-to-text translation using a pretrained speech encoder and text decoder. |
| Outcome: | The proposed approach surpasses the cascaded ST benchmark for 36 translation directions on the large-scale multilingual ST benchmark CoVoST 2. |
Neural Topic Modeling with Large Language Models in the Loop (2025.acl-long)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have demonstrated promising capabilities in topic discovery, but their direct application to topic modeling suffers from issues such as incomplete topic coverage, misalignment of topics, and inefficiency. |
| Approach: | They propose a novel LLM-in-the-loop framework that integrates Large Language Models with Neural Topic Models (NTMs) global topics and document representations are learned through the NTM, while an LLM refines these topics using an Optimal Transport (OT)-based alignment objective. |
| Outcome: | The proposed framework improves topic interpretability while preserving the efficiency of existing NTMs. |
CachePrune: Teaching LLMs What Not to Follow via KV-Cache Editing (2026.acl-long)
Copied to clipboard
Rui Wang, Junda Wu, Yu Xia, Tong Yu, Ruiyi Zhang, Ryan A. Rossi, Subrata Mitra, Lina Yao, Julian McAuley
| Challenge: | Existing Large Language Models exhibit critical vulnerability to indirect prompt injection attacks, where instructions injected within in the prompt context can override the user's intent. |
| Approach: | They propose a neural pruning algorithm that prunes neurons associated with instruction-following during KV cache encoding of the prompt context. |
| Outcome: | The proposed approach significantly reduces the attack success rate while preserving the model's ability to follow user instructions. |
EvoEdit: Evolving Null-space Alignment for Robust and Efficient Knowledge Editing (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to modifying large language models require continual updates to rectify outdated or erroneous knowledge. |
| Approach: | They propose a model editing strategy that mitigates catastrophic interference through sequential null-space alignment. |
| Outcome: | EvoEdit achieves better or comparable performance than prior state-of-the-art techniques with up to 3.53 speedup. |
POSITION BIAS MITIGATES POSITION BIAS: Mitigate Position Bias Through Inter-Position Knowledge Distillation (2025.emnlp-main)
Copied to clipboard
| Challenge: | Positional bias (PB) manifests as non-uniform sensitivity across contextual locations . previous studies have addressed PB by modifying the underlying architectures or employing extensive contextual awareness training. |
| Approach: | They propose a position-to-position knowledge distillation framework that leverages position-induced disparities to counteract PB. |
| Outcome: | The proposed framework reduces positional bias and improves performance on retrieval and reasoning tasks. |
Are Optimal Algorithms Still Optimal? Rethinking Sorting in LLM-Based Pairwise Ranking with Batching and Caching (2025.acl-short)
Copied to clipboard
Juan Wisznia, Cecilia Bolaños, Juan Tollo, Giovanni Franco Gabriel Marraffini, Agustín Andrés Gianolini, Noe Fabian Hsueh, Luciano Del Corro
| Challenge: | a new framework for analyzing sorting algorithms in pairwise ranking prompting (PRP) is developed to re-center the cost model around LLM inferences rather than traditional pairwise comparisons. |
| Approach: | They propose a framework for analyzing sorting algorithms in pairwise ranking prompting (PRP) they propose to re-center the cost model around LLM inferences rather than traditional pairwise comparisons. |
| Outcome: | The proposed framework encourages strategies such as batching and caching to mitigate inference costs. |
Can ChatGPT Assess Human Personalities? A General Evaluation Framework (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing studies study the virtual personalities of LLMs but rarely explore the possibility of analyzing human personalities via LLM. |
| Approach: | They propose to use Myers–Briggs Type Indicator (MBTI) tests to generate unbiased prompts and replace the subject in question statements to enable flexible queries and assessments. |
| Outcome: | The proposed framework enables LLMs to flexibly assess personalities of different groups of people. |
MemoPhishAgent: Memory-Augmented Multi-Modal LLM Agent for Phishing URL Detection (2026.acl-industry)
Copied to clipboard
| Challenge: | Traditional phishing website detection relies on static heuristics or reference lists, which lag behind rapidly evolving attacks. |
| Approach: | They propose a memory-augmented multi-modal LLM agent that leverages episodic memories to guide decisions on recurring and novel threats. |
| Outcome: | The proposed agent outperforms state-of-the-art phishing detection tools on two public datasets and improves recall by 20%. |
Efficiently Aligned Cross-Lingual Transfer Learning for Conversational Tasks using Prompt-Tuning (2024.findings-eacl)
Copied to clipboard
| Challenge: | Cross-lingual transfer of language models trained on high-resource languages such as English has been limited due to the high cost of obtaining non-English conversational data. |
| Approach: | They introduce a parallel and large-scale multilingual conversation dataset that is used for cross-lingual alignment pretraining by translating the English-only Schema-Guided Dialogue dataset into 105 other languages. |
| Outcome: | The proposed model performs well on slot-filling and intent classification tasks, and is able to perform well in other languages. |
Decomposed Trust: Privacy, Adversarial Robustness, Ethics, and Fairness in Low-Rank LLMs (2026.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have driven major advances across domains, yet their massive size hinders deployment in resource-constrained settings. |
| Approach: | They propose to compress large language models to reduce computation and memory consumption while maintaining accuracy. |
| Outcome: | The proposed algorithms preserve training data privacy but weaken the protection of personally identifiable information during conversations. |
Adapting Fake News Detection to the Era of Large Language Models (2024.findings-naacl)
Copied to clipboard
| Challenge: | a gap exists in understanding the interplay between machine-paraphrased real news, machine-generated fake news, and human-written real news . false information is easier to generate but harder to detect due to the bias of detectors against machine-generated texts . |
| Approach: | They propose a strategy to adapt fake news detectors to the era of large language models and AI-driven content creation . |
| Outcome: | The proposed detectors perform well on human-written articles but not vice versa . the proposed detector should be trained on datasets with lower machine-generated news ratio than the test set . |
From Allies to Adversaries: Manipulating LLM Tool-Calling through Adversarial Injection (2025.naacl-long)
Copied to clipboard
| Challenge: | Toolcalling has changed Large Language Model (LLM) applications by integrating external tools, but it also introduces new security vulnerabilities, particularly in the tool scheduling mechanisms of LLM, which have not been extensively studied. |
| Approach: | They propose a framework that exploits vulnerabilities in Large Language Models through adversarial tool injection to execute privacy theft, launch denial-of-service attacks, and manipulate business competition. |
| Outcome: | The proposed framework exploits vulnerabilities in LLM tool-calling systems through adversarial tool injection. |
Neural Media Bias Detection Using Distant Supervision With BABE - Bias Annotations By Experts (2021.findings-emnlp)
Copied to clipboard
| Challenge: | Existing studies on the detection and aggregation of media bias lack a gold standard data set and high context dependencies. |
| Approach: | They propose to use a data set to identify media bias by word and sentence level . they propose to train a model to detect bias-inducing sentences in news articles automatically . |
| Outcome: | The proposed model outperforms existing methods on a large corpus of labels on the word and sentence level. |
WebInject: Prompt Injection Attack to Web Agents (2025.emnlp-main)
Copied to clipboard
| Challenge: | Despite advanced capabilities of MLLMs, prompt injection attacks remain vulnerable to emerging security and safety threats. |
| Approach: | They propose a prompt injection attack that manipulates webpage environment to induce web agent to perform attacker-specified action. |
| Outcome: | The proposed attack outperforms baselines on multiple datasets and is highly effective. |
Structured Pruning Learns Compact and Accurate Models (2022.acl-long)
Copied to clipboard
| Challenge: | Pre-trained language models have high costs in terms of storage, memory, and computation time. |
| Approach: | They propose a task-specific structured pruning method CoFi which provides highly parallelizable subnetworks and matches distillation methods in both accuracy and latency. |
| Outcome: | The proposed method matches the distillation methods in accuracy and latency without resorting to unlabeled data. |
Improving Gender Fairness of Pre-Trained Language Models without Catastrophic Forgetting (2023.acl-short)
Copied to clipboard
| Challenge: | Existing studies addressing gender bias of pre-trained language models, usually build a small gender-neutral data set and conduct a second phase pre-training with such data. |
| Approach: | They propose a method to improve gender fairness of pre-trained models with less forgetting by evaluating them with general NLP tasks in GLUE. |
| Outcome: | The proposed method improves gender fairness of pre-trained models with less forgetting and performs better on GLUE by a large margin. |
Implicit n-grams Induced by Recurrence (2022.naacl-main)
Copied to clipboard
| Challenge: | Recent studies show that self-attention based models have limitations on modeling sequential transformations. |
| Approach: | They propose to extract some explainable features from trained RNNs that are reminiscent of classical n-grams features. |
| Outcome: | The proposed models can model interesting linguistic phenomena such as negation and intensification. |
Course-Correction: Safety Alignment Using Synthetic Preferences (2024.emnlp-industry)
Copied to clipboard
Rongwu Xu, Yishuo Cai, Zhenhong Zhou, Renjie Gu, Haiqin Weng, Liu Yan, Tianwei Zhang, Wei Xu, Han Qiu
| Challenge: | Recent studies show that large language models generate harmful content, but the potential for generating harmful content is an escalating concern. |
| Approach: | They propose to fine-tune LLMs with preference learning to emphasize the preference for timely course-correction by using an automated pipeline. |
| Outcome: | The proposed model improves course-correction skills without affecting general performance and resists jailbreak attacks. |
Hardware-Aware Parallel Prompt Decoding for Memory-Efficient Acceleration of LLM Inference (2025.findings-emnlp)
Copied to clipboard
Hao Mark Chen, Wayne Luk, Yiu Ka Fai Cedric, Rui Li, Konstantin Mishchenko, Stylianos Venieris, Hongxiang Fan
| Challenge: | Auto-regressive decoding of Large Language Models results in significant overheads in hardware performance . a novel parallel prompt decoding approach is proposed to overcome these limitations . |
| Approach: | They propose a parallel prompt decoding that uses a single model for speculation and verification. |
| Outcome: | The proposed approach speeds up auto-regressive decoding of large language models 2.49 times . it can be used on mobileLlama to Vicuna-13B on a wide range of benchmarks . |
DuReader_robust: A Chinese Dataset Towards Evaluating Robustness and Generalization of Machine Reading Comprehension in Real-World Applications (2021.acl-short)
Copied to clipboard
| Challenge: | In order to comprehensively verify the robustness and generalization of MRC models, we construct a real-world Chinese dataset - DuReader_robust . |
| Approach: | They introduce a real-world Chinese dataset to evaluate the robustness and generalization of MRC models from three aspects: over-sensitivity, over-stability and generalisation. |
| Outcome: | The proposed model fails to perform well on the challenge test set and may provide suggestions for future model development. |
VariErr NLI: Separating Annotation Error from Human Label Variation (2024.acl-long)
Copied to clipboard
| Challenge: | Existing work on label variation and annotation errors has focused on them in isolation. |
| Approach: | They propose a 2-round annotation procedure to separate human label variation from annotation errors by pairing valid explanations with annotators' validations. |
| Outcome: | The proposed procedure is based on the NLI task in English and contains 7,732 valid judgements on 1,933 explanations for 500 re-annotated items. |
Comparing Knowledge Sources for Open-Domain Scientific Claim Verification (2024.eacl-long)
Copied to clipboard
| Challenge: | Existing systems for fact-checking scientific claims assume that the documents containing the evidence are already provided and annotated or contained in a limited corpus. |
| Approach: | They perform an array of experiments to test the performance of open-domain claim verification systems on four datasets of biomedical and health claims in different settings. |
| Outcome: | The proposed system performs better with biomedical and health claims, while Wikipedia is more suited for everyday health concerns. |
Research Replication Prediction Using Weakly Supervised Learning (2020.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods to predict scientific claims’ replicability use only hand-extracted statistics features without utilizing research papers’ text information. |
| Approach: | They propose two weakly supervised learning approaches that use automatically extracted text information of research papers to improve the prediction accuracy of research replication using both labeled and unlabeled datasets. |
| Outcome: | The proposed methods achieve an accuracy of 75.76% over real-world datasets. |
Distilling Robustness into Natural Language Inference Models with Domain-Targeted Augmentation (2024.findings-acl)
Copied to clipboard
| Challenge: | Knowledge distillation optimises a smaller student model to behave similarly to a larger teacher model, retaining some performance benefits. |
| Approach: | They propose to augment the distillation with generated unlabelled examples that match the target distribution and upsamples data points among the training set that are similar to the target. |
| Outcome: | The proposed method outperforms previous robustness solutions on the task of natural language inference (NLI) it also improves performance on OOD domains even beyond the target domain. |
How Interpretable are Reasoning Explanations from Prompting Large Language Models? (2024.findings-naacl)
Copied to clipboard
| Challenge: | Prompt Engineering has garnered significant attention for enhancing the performance of large language models across a multitude of tasks. |
| Approach: | They propose a simple prompting technique that yields more than 70% improvement in interpretability. |
| Outcome: | The proposed method improves interpretability by 70% across multiple dimensions. |
DEGREE: A Data-Efficient Generation-Based Event Extraction Model (2022.naacl-main)
Copied to clipboard
I-Hung Hsu, Kuan-Hao Huang, Elizabeth Boschee, Scott Miller, Prem Natarajan, Kai-Wei Chang, Nanyun Peng
| Challenge: | Existing models for event extraction require expensive human annotations. |
| Approach: | They propose a data-efficient event extraction model that formulates event extraction as a conditional generation problem. |
| Outcome: | The proposed model can be trained with only a few labeled examples. |
Prompt-Level Distillation: A Non-Parametric Alternative to Model Fine-Tuning for Efficient Reasoning (2026.acl-industry)
Copied to clipboard
| Challenge: | Advanced reasoning typically requires Chain-of-Thought prompting, which is accurate but incurs prohibitive latency and substantial test-time inference costs. |
| Approach: | They propose to extract explicit reasoning patterns from a Teacher model and organize them into a structured list of expressive instructions for the Student model’s System Prompt. |
| Outcome: | Evaluated using Gemma-3 4B, the proposed model improves Macro F1 scores on StereoSet and Contract-NLI while increasing LogiQA accuracy to 70%. |
Predicting the Type and Target of Offensive Posts in Social Media (N19-1)
Copied to clipboard
| Challenge: | Prior work focused on detecting specific types of offensive content, such as hate speech, cyberbullying, or cyber-aggression. |
| Approach: | They propose to use a dataset to identify offensive content in social media . they compare the performance of different machine learning models to OLID . |
| Outcome: | The proposed dataset contains tweets annotated for offensive content using a fine-grained three-layer annotation scheme. |
DIA-HARM: Dialectal Disparities in Harmful Content Detection Across 50 English Dialects (2026.acl-long)
Copied to clipboard
| Challenge: | Current disinformation detection systems are predominantly developed and evaluated on Standard American English (SAE) . however, their robustness to dialectal variation is unexplored. |
| Approach: | They propose a benchmark for evaluating disinformation detection robustness across 50 English dialects . they use multi-value's linguistically-grounded transformations to introduce D-CUBE (Dialectal Disinformation Detection Corpus) |
| Outcome: | The proposed model outperforms zero-shot LLMs in human-written dialects while AI-generated content remains stable. |
Rethinking Machine Ethics – Can LLMs Perform Moral Reasoning through the Lens of Moral Theories? (2024.findings-naacl)
Copied to clipboard
| Challenge: | Existing approaches to making moral judgments are mostly bottom-up and lack explainability. |
| Approach: | They propose a top-down framework to steer Large Language Models to perform moral reasoning with well-established moral theories. |
| Outcome: | The proposed framework can integrate various moral theories on moral datasets. |
Bad Seeds: Evaluating Lexical Methods for Bias Measurement (2021.acl-long)
Copied to clipboard
| Challenge: | Existing methods for measuring bias use crowd-sourced seed lexicons, but there is little guidance for their selection. |
| Approach: | They use lexicons of different types of social biases and linguistic features to enumerate biased seeds from three English-language corpora. |
| Outcome: | The results show that seed lexicons can be used to measure bias in English-language corpora . the results show the seeds can be re-used in other contexts . |
Stigma Annotation Scheme and Stigmatized Language Detection in Health-Care Discussions on Social Media (2020.lrec-1)
Copied to clipboard
| Challenge: | a large amount of research has been done on the interpretation and influence of stigma on human behaviour and health. |
| Approach: | They develop an annotation scheme and improve the annotation process for stigma identification . they aim to distinguish stigmatised language from non-stigmatised using machine learning and NLP . |
| Outcome: | The proposed method improves the annotation process for stigma identification . the results show that the method performs better than other models . |
CAT: Confidence-Adaptive Thinking for Efficient Reasoning of Large Reasoning Models (2026.acl-industry)
Copied to clipboard
| Challenge: | Existing compression methods for large reasoning models rely on uniform length reduction or coarse-grained difficulty estimation, often leading to performance degradation on difficult problems. |
| Approach: | They propose a framework that incorporates model’s intrinsic self-certainty signals as confidence into the preference optimization process, which autonomously modulates reasoning lengths based on problem difficulty. |
| Outcome: | The proposed framework outperforms state-of-the-art models on reasoning accuracy across multiple benchmarks on different base models. |
Speaking at the Right Level: Literacy-Controlled Counterspeech Generation with RAG-RL (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing approaches produce uniform responses, ignoring that health literacy levels affect the accessibility and effectiveness of counterspeech. |
| Approach: | They propose a Controlled-Literacy framework that generates counterspeech adapted to different health literacy levels. |
| Outcome: | The proposed framework outperforms baselines by generating more accessible counterspeech to health misinformation. |
Fairness-aware Class Imbalanced Learning (2021.emnlp-main)
Copied to clipboard
| Challenge: | Existing studies on class imbalance and mitigating bias have focused on the latter . a skewed class distribution hurts the performance of deep learning models, and is often referred to as "stereotyping" |
| Approach: | They propose to extend a margin-loss based approach to enforce fairness by using tweet sentiment and occupation classification to mitigate class imbalance and demographic bias. |
| Outcome: | The proposed methods help mitigate class imbalance and demographic biases through controlled experiments. |
We Need To Talk About Random Splits (2021.eacl-main)
Copied to clipboard
| Challenge: | argued that random splits, like standard splits lead to overly optimistic performance estimates. |
| Approach: | They argue that random splits, like standard splits lead to overly optimistic performance estimates. |
| Outcome: | The proposed method leads to more realistic performance estimates than standard splits. |
PivotAttack: Rethinking the Search Trajectory in Hard-Label Text Attacks via Pivot Words (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing hard-label text attacks rely on inefficient "outside-in" strategies that traverse vast search spaces. |
| Approach: | They propose a query-efficient "inside-out" framework that perturbs Pivot Sets to induce label flips. |
| Outcome: | The proposed framework outperforms state-of-the-art methods in both Attack Success Rate and query efficiency. |
Through the MUD: A Multi-Defendant Charge Prediction Benchmark with Linked Crime Elements (2024.acl-long)
Copied to clipboard
| Challenge: | Existing charge prediction datasets focus on single-defendant cases, but real-world cases involve multiple defendants. |
| Approach: | They propose a benchmark that encompasses legal cases involving multiple defendants . they develop an interpretable model called EJudge that incorporates crime elements and legal rules to infer charges. |
| Outcome: | The proposed model outperforms state-of-the-art models in predicting crime charges while providing corresponding rationales. |
Efficient and Robust Question Answering from Minimal Context over Documents (P18-1)
Copied to clipboard
| Challenge: | Recent work shows that neural QA models are sensitive to adversarial inputs. |
| Approach: | They propose a sentence selector to select the minimal set of sentences to feed into a QA model. |
| Outcome: | The proposed system reduces training time and inference time by up to 13 times . it is comparable to or better than the state-of-the-art on SQuAD, NewsQA, TriviaQA and SQu AD-Open . |
Unsupervised Learning of Syntactic Structure with Invertible Neural Projections (D18-1)
Copied to clipboard
| Challenge: | Unsupervised learning of syntactic structure is typically performed using generative models with discrete latent variables and multinomial parameters. |
| Approach: | They propose a generative model that jointly learns discrete syntactic structure and continuous word representations in an unsupervised fashion by cascading an invertible neural network with a structured generative prior. |
| Outcome: | The proposed model outperforms state-of-the-art models on part-of speech (POS) induction and unsupervised dependency parsing without gold POS annotation. |
Towards Reward Fairness in RLHF: From a Resource Allocation Perspective (2025.acl-long)
Copied to clipboard
| Challenge: | if rewards are imperfect, they can adversely affect the alignment of large language models (LLMs). |
| Approach: | They propose a bias-agnostic method to address the issue of reward unfairness from a resource allocation perspective without specifically designing for each type of bias . they apply methods Fairness Regularization and Fairness Coefficient to achieve fairness in rewards. |
| Outcome: | The proposed method achieves fairness in rewards while minimizing biases . it can be applied to verification and reinforcement learning scenarios . |
Alignment-Enhanced Decoding: Defending Jailbreaks via Token-Level Adaptive Refining of Probability Distributions (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing defenses against jailbreaks focus on perturbing or inspecting inputs, but ignore competing objectives, the underlying cause of alignment failures. |
| Approach: | They propose a novel defense that employs adaptive decoding to address the root causes of jailbreak issues. |
| Outcome: | The proposed defense improves safety alignment while maintaining helpfulness. |
Changing the World by Changing the Data (2021.acl-long)
Copied to clipboard
| Challenge: | a new paper argues that data curation is already happening, and it is changing the world . social biases and spurious patterns are attracting more attention in NLP models . |
| Approach: | They argue that data curation is already happening and will be happening . they argue that social biases and spurious patterns are the main problems . |
| Outcome: | a new paper argues that data curation is already and will be happening, and it is changing the world. |
Self-Supervised Position Debiasing for Large Language Models (2024.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for debiasing large language models require external bias knowledge or annotated non-biased samples, which is lacking for position debiases. |
| Approach: | They propose a self-supervised position debiasing framework that leverages unsupervised responses from pre-trained LLMs for debiazing without external bias knowledge. |
| Outcome: | The proposed framework outperforms existing methods in mitigating three types of position biases on eight datasets and five tasks. |
HateModerate: Testing Hate Speech Detectors against Content Moderation Policies (2024.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies on hate speech detection have failed to answer this question. |
| Approach: | They propose a dataset for testing the behaviors of automated content moderators against content policies. |
| Outcome: | The proposed dataset includes hateful and non-hateful examples matching the 41 community standards guideline policies of Facebook. |
Calibration Meets Explanation: A Simple and Effective Approach for Model Confidence Estimates (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods to improve confidence calibration of pre-trained language models are still a mystery. |
| Approach: | They propose a method that leverages model explanations to make models less confident with non-inductive attributions. |
| Outcome: | The proposed method improves confidence calibration in all settings and reduces calibration errors when combined with temperature scaling. |
PVPO: Pre-Estimated Value-Based Policy Optimization for Agentic Reasoning (2026.findings-acl)
Copied to clipboard
| Challenge: | extending grouping-based methods to agentic reasoning presents unique challenges . frequent environment interactions and tool invocations render intra-group advantage estimation unstable . |
| Approach: | They propose a grouping-based method that uses a single round of rollouts to stabilize advantage estimation. |
| Outcome: | a new RL framework outperforms grouping-based methods in retrieval tasks and advanced mathematical reasoning benchmarks. |
Collaboration or Corporate Capture? Quantifying NLP’s Reliance on Industry Artifacts and Contributions (2024.acl-long)
Copied to clipboard
| Challenge: | EMNLP 2022 citations are three times greater than expected for pre-trained models . industry participation in the Association of Computational Linguistics (ACL) anthology has increased 180% from 2017 to 2022. |
| Approach: | They surveyed 100 papers published at EMNLP 2022 to determine the ratio of their citations to industry models. |
| Outcome: | a new study shows that industry citations are three times greater than expected . the study aims to better understand whether industry collaboration is still collaboration . industry participation in the 2023 AI index report is the top takeaway . |
Generating Data to Mitigate Spurious Correlations in Natural Language Inference Datasets (2022.acl-long)
Copied to clipboard
| Challenge: | Natural language processing models exploit spurious correlations between features and labels in datasets to perform well only within the distributions they are trained on. |
| Approach: | They propose to generate a debiased version of a dataset and replace it with training data to train a model that is generalised to different task distributions. |
| Outcome: | The proposed method outperforms or performs comparable to state-of-the-art debiasing strategies on a large suite of debiased, out-of distribution, and adversarial test sets. |
Towards A Friendly Online Community: An Unsupervised Style Transfer Framework for Profanity Redaction (2020.coling-main)
Copied to clipboard
| Challenge: | Existing methods for redacting offensive comments into non-offensive ones are inadequate to detect hateful content on social media platforms. |
| Approach: | They propose a method for transforming offensive comments into non-offensive ones using a Retrieve, Generate and Edit unsupervised style transfer pipeline. |
| Outcome: | The proposed method outperforms existing models on automatic metrics and human evaluations and consistently performs well on all automatic evaluation metrics. |
BioRo: The Biomedical Corpus for the Romanian Language (L18-1)
Copied to clipboard
| Challenge: | Biomedical text mining uses linguistic resources available in English, but for other languages such as Romanian, the access to language resources is not straight-forward. |
| Approach: | They present a biomedical corpus of the Romanian language, which is a valuable linguistic asset for biomedically text mining. |
| Outcome: | The proposed corpus will be made publicly available to the biomedical text mining community . the corpus is a reference corpus for the Romanian language . |
Evaluating Debiasing Techniques for Intersectional Biases (2021.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for debiasing protected attributes have been limited to binary attributes in isolation, however many corpora involve multiple such attributes, possibly with higher cardinality. |
| Approach: | They propose to evaluate a bias-constrained model which is new to NLP and an extension of the iterative nullspace projection technique which can handle multiple identities. |
| Outcome: | The proposed model is based on a new iterative nullspace projection technique which can handle multiple identities. |
Reducing Spurious Correlations in Aspect-based Sentiment Analysis with Explanation from Large Language Models (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Aspect-based sentiment analysis models are susceptible to learning spurious correlations between words . a recent study shows that feature engineering is time-consuming and costly . |
| Approach: | They propose to use a template to prompt LLMs to generate an appropriate explanation for the sentiment polarity of each aspect to reduce spurious correlations. |
| Outcome: | The proposed methods improve ABSA models and their generalization ability. |
The Two Paradigms of LLM Detection: Authorship Attribution vs Authorship Verification (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for detecting texts generated by large language models are disputed . authors argue that there are limitations in the current technology . |
| Approach: | They propose to make LLM detectors robust against domain shifts and build benchmarks . they argue that the limitations lie elsewhere, and open the realm of authorship analysis technology . |
| Outcome: | The proposed method systematically analyzes the benchmarks and validates it using state-of-the-art detectors. |
BitBypass: A New Direction in Jailbreaking Aligned Large Language Models with Bitstream Camouflage (2026.findings-eacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have inherent risk of generating harmful and unsafe content. |
| Approach: | They develop a black-box jailbreak attack that leverages hyphen-separated bitstream camouflage to bypass aligned Large Language Models' safety alignment. |
| Outcome: | The proposed attack outperforms state-of-the-art jailbreak attacks in stealthiness and attack success. |
Decorrelate Irrelevant, Purify Relevant: Overcome Textual Spurious Correlations from a Feature Perspective (2022.coling-1)
Copied to clipboard
| Challenge: | Existing methods to debiase samples with biased features obstructs the model in learning from non-biased parts of the samples. |
| Approach: | They propose to eliminate spurious correlations in a fine-grained manner from a feature space perspective by using Random Fourier Features and weighted re-sampling to decorrelate dependencies between features. |
| Outcome: | The proposed method eliminates spurious correlations in a fine-grained manner from a feature space perspective. |
Parameter-efficient Modularised Bias Mitigation via AdapterFusion (2023.eacl-main)
Copied to clipboard
Deepak Kumar, Oleg Lesota, George Zerveas, Daniel Cohen, Carsten Eickhoff, Markus Schedl, Navid Rekabsaz
| Challenge: | Large pre-trained language models contain societal biases and carry along these biase . Current approaches to mitigate these bias impose debiasing by updating model parameters, effectively transferring model to irreversible debiased state. |
| Approach: | They propose to develop stand-alone debiasing functionalities separate from the model, which can be integrated into the model on-demand while keeping the core model untouched. |
| Outcome: | The proposed approach improves or maintains effectiveness of bias mitigation, avoids catastrophic forgetting in a multi-attribute scenario, and maintains on-par task performance while granting parameter-efficiency and easy switching between the original and debiased models. |
A Girl Has A Name: Detecting Authorship Obfuscation (2020.acl-main)
Copied to clipboard
| Challenge: | Existing authorship attribution methods are not stealthy as they degrade text smoothness in detectable manner. |
| Approach: | They evaluate the stealthiness of authorship attribution methods under an adversarial threat model and show that they are not stealthy . |
| Outcome: | The proposed methods can be identified with an average F1 score of 0.87 . |
Mutual-pairing Data Augmentation for Fewshot Continual Relation Extraction (2025.naacl-long)
Copied to clipboard
Nguyen Hoang Anh, Quyen Tran, Thanh Xuan Nguyen, Nguyen Thi Ngoc Diep, Linh Ngo Van, Thien Huu Nguyen, Trung Le
| Challenge: | Existing methods for Few-shot Continual Relation Extraction struggle with catastrophic forgetting and overfitting. |
| Approach: | They propose a method that transforms single input sentences into complex texts by integrating old and new data. |
| Outcome: | The proposed method sharpens model focus and improves model performance . it also uncovers fascinating behaviors of Sharpness-Aware Minimization (SAM) in Few-shot Continual Learning. |
A Zero-Shot Monolingual Dual Stage Information Retrieval System for Spanish Biomedical Systematic Literature Reviews (2024.naacl-long)
Copied to clipboard
| Challenge: | Existing studies have shown that most SRs are skewed towards English databases, excluding databases in Languages other than English (LoE). |
| Approach: | They propose a zero-shot dual information retrieval baseline system that integrates traditional retrieval methods with pre-trained language models and cross-attention re-rankers for enhanced accuracy in Spanish biomedical literature retrieval. |
| Outcome: | The proposed system improves on three real-life case studies in Spanish biomedical literature retrieval using the LILACS database, which is known for its coverage of Latin American and Caribbean biomedically literature. |
An End-to-End Submodular Framework for Data-Efficient In-Context Learning (2024.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in natural language tasks leverage the emergent In-Context Learning ability of pretrained Large Language Models (LLMs). |
| Approach: | They propose a framework for exemplar selection for in-context learning that uses a pool-based active learning approach to select Diverse and informative exemplars from the target tasks’ unlabeled pool. |
| Outcome: | The proposed framework outperforms existing methods for data annotation and similarity-based methods for test query-specific exemplar retrieval on 7 different NLP datasets and 5 LLMs of varying complexities. |
Toxicity Detection is NOT all you Need: Measuring the Gaps to Supporting Volunteer Content Moderators through a User-Centric Method (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing efforts to automate content moderation have focused on identifying toxic, offensive, and hateful content . yet, it remains unclear whether improvements have addressed the needs of volunteer content moderators . |
| Approach: | They propose to use a model review to examine the availability of moderators' models to flag violations of various forum rules. |
| Outcome: | The proposed models perform poorly on a significant portion of the rules. |
At the Intersection of NLP and Sustainable Development: Exploring the Impact of Demographic-Aware Text Representations in Modeling Value on a Corpus of Interviews (2022.lrec-1)
Copied to clipboard
| Challenge: | In order to preserve the privacy of speakers, we investigate encoding demographic information using autoencoders. |
| Approach: | They introduce a dataset of qualitative interviews from rural communities in India and Uganda and use it to enhance text representations with demographic information. |
| Outcome: | The proposed model extends the UPV classification model with demographic information to preserve the privacy of speakers. |
ImaRA: An Imaginative Frame Augmented Method for Low-Resource Multimodal Metaphor Detection and Explanation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for multimodal metaphor detection neglect cross-domain and attribute similarity characteristics underlying multimodal understanding. |
| Approach: | They propose an Imaginative FRame Augmented method for multimodal metaphor detection and explanation . they use a cross-modal imagination dataset rich in multimodal multimodal expressions . |
| Outcome: | The proposed method outperforms existing methods with training data on two datasets. |
Why Attention is Not Explanation: Surgical Intervention and Causal Reasoning about Neural Models (2020.lrec-1)
Copied to clipboard
| Challenge: | a recent study finds brittleness in explanations obtained through attention mechanisms . a philosophy of science theory allows robust yet non-causal reasoning in explanation . |
| Approach: | They propose to use philosophy of science to examine the state-of-the-art in explanation for NLP models . they argue that it is impossible to explain attention-based learning by attention mechanisms . |
| Outcome: | The proposed model selection criteria are based on philosophy of science theories . the proposed model is based upon a model that is more explainable than a classical model . |
Features or Spurious Artifacts? Data-centric Baselines for Fair and Robust Hate Speech Detection (2022.naacl-main)
Copied to clipboard
| Challenge: | lexical biases in hate speech detection are limited when applied to real-world data, exhibiting limited out-of-distribution robustness and perpetuating harmful social biase. |
| Approach: | They propose to disentangle spurious and authentic artifacts and analyze their impact on out-of-distribution fairness and robustness. |
| Outcome: | The proposed models show that spurious artifacts require different treatments to attain robustness and fairness in hate speech detection. |
DisCup: Discriminator Cooperative Unlikelihood Prompt-tuning for Controllable Text Generation (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing prompt tuning approaches for attribute-controllable text generation are difficult to implement due to the lack of interpretability of deep neural networks. |
| Approach: | They propose a new approach that incorporates attribute knowledge of discriminator to optimize prompt tuning by steering a frozen CLM to produce attribute-specific texts. |
| Outcome: | The proposed approach can achieve state-of-the-art control performance while maintaining high-quality text generation. |
CONSCENDI: A Contrastive and Scenario-Guided Distillation Approach to Guardrail Models for Virtual Assistants (2024.naacl-long)
Copied to clipboard
| Challenge: | A major challenge in deploying LLM-based virtual conversational assistants in real world settings is ensuring they operate within what is admissible for the task. |
| Approach: | They propose to use large language models (LLMs) to generate training data with two key LLM components: scenario-augmented generation and contrastive training examples. |
| Outcome: | The proposed model improves over baselines in multiple dialogue domains. |
Language Model-Driven Data Pruning Enables Efficient Active Learning (2026.findings-eacl)
Copied to clipboard
| Challenge: | Existing data pruning methods for active learning are expensive and time-consuming. |
| Approach: | They propose a plug-and-play data pruning strategy that leverages language models to prune the unlabeled pool. |
| Outcome: | The proposed pruning strategy outperforms existing pruning methods on translation, sentiment analysis, topic classification, and summarization tasks on diverse datasets. |
Boundary-Aware LLM Augmentation for Low-Resource Event Argument Extraction (2026.eacl-long)
Copied to clipboard
| Challenge: | Event argument extraction (EAE) is a crucial task in information extraction but its performance heavily depends on expensive annotated data. |
| Approach: | They investigate argument replacement, adjunction rewriting, their combination, and annotation generation using four LLM-based augmentation strategies. |
| Outcome: | The proposed methods improve performance over boundary-agnostic methods and provide detailed analysis of quality from multiple perspectives. |
Building a Data Infrastructure for a Mid-Resource Language: The Case of Catalan (2024.lrec-main)
Copied to clipboard
Aitor Gonzalez-Agirre, Montserrat Marimon, Carlos Rodriguez-Penagos, Javier Aula-Blasco, Irene Baucells, Carme Armentano-Oller, Jorge Palomar-Giner, Baybars Kulebi, Marta Villegas
| Challenge: | Aina Project aims to provide Catalan with the resources needed to keep its relevance in AI/NLP applications. |
| Approach: | They propose a set of strategies to consider when improving technology support for a mid- or low-resource language . they propose annotated datasets and a framework to make models ready to use . |
| Outcome: | The Aina Project aims to provide Catalan with the necessary resources to keep its relevance in AI/NLP-related industry and research. |
Unsupervised Extractive Summarization-Based Representations for Accurate and Explainable Collaborative Filtering (2021.acl-long)
Copied to clipboard
| Challenge: | Existing extractive summarization-based collaborative filtering models learn accurate representations of users and items based on user-given numeric ratings, but employing them is an oversimplification of user preferences and item characteristics. |
| Approach: | They propose to use BERT, K-Means embedding clustering, and multilayer perceptron to learn sentence embeddations, representation-explanations, and user-item interactions to create extractive summaries. |
| Outcome: | The proposed model improves rating prediction accuracy and user/item explainability. |
Uncertainty-aware Propagation Structure Reconstruction for Fake News Detection (2022.coling-1)
Copied to clipboard
| Challenge: | Existing methods to detect fake news neglect a broader propagation uncertainty issue . Existing studies leverage the user interactions in a social media conversation thread to detect false news. |
| Approach: | They propose a dual graph-based model for improving fake news detection . they propose to explore latent interactions in the actual propagation . |
| Outcome: | The proposed model improves on two real-world datasets showing that it is superior to existing models. |
Explanation based In-Context Demonstrations Retrieval for Multilingual Grammatical Error Correction (2025.naacl-long)
Copied to clipboard
| Challenge: | Grammatical error correction (GEC) aims to correct grammatical, spelling, and semantic errors in natural language text. |
| Approach: | They propose a retrieval method based on natural language grammatical error explanations to match inputs with pre-constructed databases where explanations for erroneous samples are generated by LLMs. |
| Outcome: | The proposed method outperforms existing semantic and BM25-based retrieval techniques without additional training or language adaptation. |
Adversarial Domain Adaptation for Machine Reading Comprehension (D19-1)
Copied to clipboard
| Challenge: | Existing models for machine reading comprehension rely on large amounts of human-annotated in-domain data. |
| Approach: | They propose an unsupervised domain adaptation framework for Machine Reading Comprehension where the source domain has a large amount of labeled data, while only unlabeled passages are available in the target domain. |
| Outcome: | The proposed framework can be generalizable to different MRC models and datasets and can be extended to semi-supervised learning. |
Flames: Benchmarking Value Alignment of LLMs in Chinese (2024.naacl-long)
Copied to clipboard
Kexin Huang, Xiangyang Liu, Qianyu Guo, Tianxiang Sun, Jiawei Sun, Yaru Wang, Zeyang Zhou, Yixu Wang, Yan Teng, Xipeng Qiu, Yingchun Wang, Dahua Lin
| Challenge: | Existing benchmarks for large language models (LLMs) do not accurately uncover safety vulnerabilities in LLMs. |
| Approach: | They propose a value alignment benchmark called Flames that encompasses both harmlessness principles and a unique morality dimension that integrates specific Chinese values such as harmony. |
| Outcome: | The proposed model performs poorly on Flames, particularly in safety and fairness dimensions. |
Analytic Automated Essay Scoring Based on Deep Neural Networks Integrating Multidimensional Item Response Theory (2022.coling-1)
Copied to clipboard
| Challenge: | Essay exams have two drawbacks in that grading them is expensive and raises questions about fairness. |
| Approach: | They propose to use a multidimensional item response theory model to improve interpretability while maintaining scoring accuracy. |
| Outcome: | The proposed model improves interpretability while maintaining accuracy while preserving cost and accuracy. |
Automated Progressive Red Teaming (2025.coling-main)
Copied to clipboard
| Challenge: | Automated red teaming (ART) is effective but time-consuming, costly and lacks scalability. |
| Approach: | They propose an automated red teaming framework that generates adversarial prompts to expose LLM vulnerabilities. |
| Outcome: | The proposed framework explores and exploits LLM vulnerabilities through multi-round interactions. |
WavLLM: Towards Robust and Adaptive Speech Large Language Model (2024.findings-emnlp)
Copied to clipboard
Shujie Hu, Long Zhou, Shujie Liu, Sanyuan Chen, Lingwei Meng, Hongkun Hao, Jing Pan, Xunying Liu, Jinyu Li, Sunit Sivasankaran, Linquan Liu, Furu Wei
| Challenge: | Recent advances in large language models (LLMs) have expanded their scope to encompass multimodal functions. |
| Approach: | They propose a robust and adaptive speech large language model with dual encoders . they validate the model on universal speech benchmarks and apply it to specialized speech-question-answer datasets based on a CoT approach . |
| Outcome: | The proposed model achieves state-of-the-art performance across a range of speech tasks on the same model size. |
It’s Morphin’ Time! Combating Linguistic Discrimination with Inflectional Perturbations (2020.acl-main)
Copied to clipboard
| Challenge: | Existing work on societal bias in NLP focuses on race and gender . linguistic background is a unique attribute that has been largely ignored in the field . |
| Approach: | They examine linguistic background to craft plausible adversarial examples that expose biases in popular NLP models. |
| Outcome: | The proposed model improves robustness without sacrificing performance on clean data. |
Mitigating Gender Bias Amplification in Distribution by Posterior Regularization (2020.acl-main)
Copied to clipboard
| Challenge: | Recent studies show that data-driven machine learning models carry societal biases in the dataset they trained on. |
| Approach: | They propose to calibrate top predictions of a model by injecting corpus-level constraints to ensure that the gender disparity is not amplified. |
| Outcome: | The proposed method can almost remove bias amplification in the distribution with little loss of performance. |
CDA: a Cost Efficient Content-based Multilingual Web Document Aligner (2021.eacl-main)
Copied to clipboard
| Challenge: | a Content-based document alignment approach is an efficient way to align multilingual web documents based on content. |
| Approach: | They propose a Content-based document alignment approach to align multilingual web documents based on content in parallel training data for machine translation systems. |
| Outcome: | The proposed method achieves comparable performance with state-of-the-art systems in the WMT-16 Bilingual Document Alignment Shared Task benchmark while operating in multilingual space. |
CoRanking: Collaborative Ranking with Small and Large Ranking Agents (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Listwise ranking based on Large Language Models (LLMs) has achieved state-of-the-art performance in Information Retrieval (IR) however, their effectiveness often depends on LLMs with massive parameter scales and computationally expensive sliding window processing, leading to substantial efficiency bottlenecks. |
| Approach: | They propose a Collaborative Ranking framework (CoRanking) for LLM-based listwise ranking based on large language models with massive parameter scales and computationally expensive sliding window processing. |
| Outcome: | The proposed framework reduces ranking latency by approximately 70% while improving effectiveness compared to the standalone large reranker. |
Look Twice before You Leap: A Rational Framework for Localized Adversarial Text Anonymization (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing LLMs rely on remote API services, which creates privacy paradoxes and suboptimal solutions with severe utility collapse. |
| Approach: | They propose a localized and training-free framework with an Attacker-Arbitrator-Anonymizer architecture that allows attackers to filter out ghost leaks. |
| Outcome: | The proposed framework achieves superior privacy-utility trade-off compared to strong baselines. |
OCP: Outlier-Centric Probing for Dynamic Structured Pruning of LLMs (2026.acl-long)
Copied to clipboard
| Challenge: | Existing structured pruning methods fail to identify outlier-triggering tokens and uniform layer-wise sparsity misaligns with heterogeneous outlier distributions. |
| Approach: | They propose a framework that prioritizes capturing outlier-triggering tokens rather than reconstructing full hidden distributions. |
| Outcome: | Experiments on LLaMA2, LLama3 and OPT show that the proposed framework outperforms state-of-the-art methods and achieves 25% perplexity reduction at 1.6 speedup. |
Towards Efficient and Robust VQA-NLE Data Generation with Large Vision-Language Models (2025.coling-main)
Copied to clipboard
| Challenge: | Existing methods for creating a vision question-answering with natural language explanations rely on human annotations that are time-consuming and costly. |
| Approach: | They propose a method that generates high-quality natural language explanations using LVLMs by using visual prompts. |
| Outcome: | The proposed method generates high-quality synthetic VQA-NLE datasets 20x faster than human annotations with minimal decrease in qualitative metrics. |
DefVerify: Do Hate Speech Models Reflect Their Dataset’s Definition? (2025.coling-main)
Copied to clipboard
| Challenge: | DefVerify is a 3-step procedure that encodes a user-specified definition of hate speech, quantifies to what extent the model reflects the intended definition, and identifies the point of failure in the workflow. |
| Approach: | They propose a 3-step procedure that encodes a user-specified definition of hate speech and quantifies to what extent the model reflects intended definition. |
| Outcome: | The proposed procedure detects gaps between definition and model behavior when applied to six popular hate speech benchmark datasets. |
On the Impact of Temporal Concept Drift on Model Explanations (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Explanation faithfulness of model predictions is typically evaluated on held-out data from the same temporal distribution as the training data. |
| Approach: | They examine the impact of temporal variation on model explanations extracted by eight feature attribution methods and three select-then-predict models across six text classification tasks. |
| Outcome: | The proposed method shows the most robust faithfulness scores across datasets and in asynchronous settings. |
FiD-Ex: Improving Sequence-to-Sequence Models for Extractive Rationale Generation (2021.emnlp-main)
Copied to clipboard
| Challenge: | Pre-trained sequence to sequence models are effective in making and generating NL explanations, but they have many shortcomings. |
| Approach: | They propose a model that uses sentence markers to eliminate explanation fabrication . they use fusion-in-decoder architecture to handle long input contexts . |
| Outcome: | The proposed model significantly improves on the ERASER explainability benchmark. |
Ignore This Title and HackAPrompt: Exposing Systemic Vulnerabilities of LLMs Through a Global Prompt Hacking Competition (2023.emnlp-main)
Copied to clipboard
Sander Schulhoff, Jeremy Pinto, Anaum Khan, Louis-François Bouchard, Chenglei Si, Svetlina Anati, Valen Tagliabue, Anson Kost, Christopher Carnahan, Jordan Boyd-Graber
| Challenge: | Large Language Models are increasingly being deployed in interactive contexts that involve direct user engagement. |
| Approach: | They run a global prompt hacking competition to encourage research on prompt hacks . they elicit 600K+ adversarial prompts against three state-of-the-art LLMs based on a dataset . |
| Outcome: | The results of the competition show that current LLMs can be manipulated via prompt hacking . the competition elicits 600K+ adversarial prompts against three state-of-the-art LLM models . |
VideoEraser: Concept Erasure in Text-to-Video Diffusion Models (2025.emnlp-main)
Copied to clipboard
| Challenge: | Experimental results show that VideoEraser outperforms prior methods regarding efficacy, integrity, fidelity, robustness, and generalizability. |
| Approach: | They propose a training-free framework that prevents T2V diffusion models from generating videos with undesirable concepts even when explicitly prompted with those concepts. |
| Outcome: | The proposed framework outperforms existing methods in erasure, celebrity erasion, and explicit content erasing tasks. |
DeAR: Dual-Stage Document Reranking with Reasoning Agents via LLM Distillation (2025.findings-emnlp)
Copied to clipboard
| Challenge: | DeAR is an open-source framework that decouples the tasks of LLMs with holistic cross-document analysis. |
| Approach: | They propose an open-source framework that decouples relevance scoring with holistic cross-document analysis. |
| Outcome: | The proposed framework outperforms open-source frameworks in QA and open-domain QA. |
Not Far Away, Not So Close: Sample Efficient Nearest Neighbour Data Augmentation via MiniMax (2021.findings-acl)
Copied to clipboard
| Challenge: | Existing kNN-based augmentation techniques blindly incorporate all samples, but MiniMax-kNN uses a subset of augmented samples to maximize KL-divergence between teacher and student models. |
| Approach: | They propose a semi-supervised approach to augmented data augmentation using kNN. |
| Outcome: | The proposed method outperforms existing kNN-based augmentation techniques on several classification tasks and requires fewer augmented examples and less computation to achieve superior performance. |
ProvBench: A Benchmark of Legal Provision Recommendation for Contract Auto-Reviewing (2025.acl-long)
Copied to clipboard
Xiuxuan Shen, Zhongyuan Jiang, Junsan Zhang, Junxiao Han, Yao Wan, Chengjie Guo, Bingcheng Liu, Jie Wu, Renxiang Li, Philip S. Yu
| Challenge: | Contract review is labor-intensive, time-consuming, and costly . a benchmark is proposed to detect potential legal conflicts . |
| Approach: | They propose a benchmark for legal provision recommendation and conflict detection for contract auto-reviewing which aims to recommend the legal provisions related to contract clauses and detect possible legal conflicts. |
| Outcome: | The proposed task recommends legal provisions related to contract clauses and detects legal conflicts. |
Event-Radar: Event-driven Multi-View Learning for Multimodal Fake News Detection (2024.acl-long)
Copied to clipboard
| Challenge: | Existing methods for detecting multimedia fake news have demonstrated excellent results . however, addressing event-level inconsistency and learning from poor-quality news remains a challenge . |
| Approach: | They propose an Event-diven fake news detection framework that integrates visual manipulation, textual emotion and multimodal inconsistency at event-level for fake news identification. |
| Outcome: | The proposed framework performs well on three large-scale fake news detection benchmarks. |
The Relevance of Value Systems for Offensive Language Detection (2026.eacl-long)
Copied to clipboard
| Challenge: | Recent research in perspectivism has departed from the assumption that offensiveness can be defined through a universal perspective. |
| Approach: | They propose to use a dataset consisting of neutrally-phrased sentences on controversial topics, evaluated by individuals from 4 different value systems to identify offensiveness patterns. |
| Outcome: | The proposed dataset consists of neutrally-phrased sentences on controversial topics, evaluated by individuals from 4 different value systems. |
Towards Robust and Generalized Parameter-Efficient Fine-Tuning for Noisy Label Learning (2024.acl-long)
Copied to clipboard
| Challenge: | Parameter-efficient fine-tuning (PEFT) has enabled efficient optimization of cumbersome language models in real-world environments. |
| Approach: | They propose a routing-based PEFT approach that adaptively activates PEFT modules. |
| Outcome: | The proposed method is more sensitive to noise interference than other methods. |
How low is too low? A monolingual take on lemmatisation in Indian languages (2021.naacl-main)
Copied to clipboard
| Challenge: | Prior work on ML based lemmatization focused on high resource languages, where data sets (word forms) are readily available. |
| Approach: | They propose to use neural methods to relate inflected forms of words to their dictionary form to reduce the sparse data problem. |
| Outcome: | The proposed methods can give competitive accuracy even in low resource setting. |
Thinking Outside of the Differential Privacy Box: A Case Study in Text Privatization with Language Model Prompting (2024.emnlp-main)
Copied to clipboard
| Challenge: | Recent studies have focused on the integration of Differential Privacy (DP) into NLP techniques. |
| Approach: | They propose a method for text privatization leveraging language models to rewrite texts . they examine the usability of DP in NLP and its benefits over non-DP approaches . |
| Outcome: | The proposed method is a novel method for text privatization leveraging language models to rewrite texts. |
Tuning-Free Personalized Alignment via Trial-Error-Explain In-Context Learning (2025.findings-naacl)
Copied to clipboard
Hyundong Justin Cho, Karishma Sharma, Nicolaas Paul Jedema, Leonardo F. R. Ribeiro, Jonathan May, Alessandro Moschitti
| Challenge: | Language models are biased towards generic outputs as they are trained to align to an aggregate preference to be generally useful. |
| Approach: | They propose a tuning-free method that personalizes language models for text generation tasks with fewer than 10 examples per user. |
| Outcome: | The proposed method achieves favorable win rates on pairwise comparisons with the previous state-of-the-art and outperforms competitive tuning-free baselines for personalized alignment tasks of writing emails, essays and news articles. |
Evaluating Machine Translation Datasets for Low-Web Data Languages: A Gendered Lens (2026.findings-acl)
Copied to clipboard
Hellina Hailu Nigatu, Bethelhem Yemane Mamo, Bontu Fufa Balcha, Debora Taye Tesfaye, Elbethel Daniel Zewdie, Ikram Behiru Nesiru, Jitu Ewnetu Hailu, Senait Mengesha Yayo
| Challenge: | afan oromo, amharic, and tigrinya are low-resourced languages . they are used for training, benchmarks, news, health, and sports . afono o'mara: quantity does not guarantee quality of MT datasets . |
| Approach: | They investigate the quality of machine translation datasets for three low-resourced languages . they found a large skew towards the male gender in the datasets . |
| Outcome: | The results show that training data has large representation of political and religious text, but benchmark datasets focus on news, health, and sports. |
Language model acceptability judgements are not always robust to context (2023.acl-long)
Copied to clipboard
Koustuv Sinha, Jon Gauthier, Aaron Mueller, Kanishka Misra, Keren Fuentes, Roger Levy, Adina Williams
| Challenge: | a recent study found that models prefer acceptable inputs over acceptable ones. |
| Approach: | They find that model judgements are generally robust when placed in randomly sampled linguistic contexts, but unstable when contexts match the test stimuli in syntactic structure. |
| Outcome: | The proposed model performance improves when contexts match syntactic structure, and declines when they are unacceptable. |
Dual Adversarial Neural Transfer for Low-Resource Named Entity Recognition (P19-1)
Copied to clipboard
| Challenge: | Named entity recognition (NER) is an important step in most natural language processing (NLP) applications. |
| Approach: | They propose a dual-adversarial neural transfer method for addressing low-resource Named Entity Recognition (NER) they propose 'Generalized Resource-Adversarial Discriminator' and 'accidental training' |
| Outcome: | The proposed method improves on low-resource Named Entity Recognition (NER) with two variants, i.e., DATNet-F and DATNET-P, and adversarial training is adopted to boost model generalization. |
Extracting Age-Related Stereotypes from Social Media Texts (2022.lrec-1)
Copied to clipboard
| Challenge: | a method for extracting age-related stereotypes from Twitter data is under-studied in NLP . stereotyping on the basis of protected characteristics has been understudied . |
| Approach: | They propose a method for extracting age-related stereotypes from Twitter data . they generate a corpus of 300,000 over-generalizations about four contemporary generations . |
| Outcome: | The method uncovers common stereotypes as reported in media and psychological literature . it also finds that stereotypes for different generations vary across topics . |
D-CALM: A Dynamic Clustering-based Active Learning Approach for Mitigating Bias (2023.findings-acl)
Copied to clipboard
| Challenge: | Infusing clustering with active learning with AL can overcome the bias issue of both AL and traditional annotation methods while exploiting AL’s annotation efficiency. |
| Approach: | They propose an algorithm that dynamically adjusts clustering and annotation efforts in response to an estimated classifier error-rate. |
| Outcome: | The proposed algorithm outperforms baseline AL approaches with pretrained transformers and traditional Support Vector Machines on eight datasets for emotion, hatespeech, dialog act, and book type detection tasks. |
Tracking Life’s Ups and Downs: Mining Life Events from Social Media Posts for Mental Health Analysis (2025.acl-long)
Copied to clipboard
| Challenge: | Existing studies have indicated that major life events can greatly impact individuals’ mental health, but shedding its light on social media data is challenging due to the complexity and ambiguity nature of life events. |
| Approach: | They propose to extract life events mentioned in posts on social media to uncover a social media event dataset which includes 12 major life event categories that are likely to occur in everyday life. |
| Outcome: | The proposed dataset includes 12 life event categories that are likely to occur in everyday life and is human-annotated under iterative procedure and boasts a high level of quality. |
Counterfactually Augmented Data and Unintended Bias: The Case of Sexism and Hate Speech Detection (2022.naacl-main)
Copied to clipboard
| Challenge: | sexism and hate speech detection models may be over-relying on core features . construct-driven CAD may induce models to ignore context in which core features are used . |
| Approach: | They propose to use construct-driven and construct-agnostic CAD to reduce model bias . sexism and hate speech detection models are trained on counterfactually augmented data . |
| Outcome: | Using a diverse set of CAD—construct-driven and construct-agnostic—reduces unintended bias. |
Responsible Evaluation of AI for Mental Health (2026.acl-long)
Copied to clipboard
Hiba Arnaout, Anmol Goel, H. Andrew Schwartz, Steffen T. Eberhardt, Dana Atzil-Slonim, Gavin Doherty, Brian Schwartz, Wolfgang Lutz, Tim Althoff, Munmun De Choudhury, Hamidreza Jamalabadi, Raj Sanjay Shah, Flor Miriam Plaza-del-Arco, Dirk Hovy, Maria Liakata, Iryna Gurevych
| Challenge: | Existing approaches to evaluating AI tools in this domain remain fragmented and inconsistent. |
| Approach: | They propose a taxonomy of AI mental health support types that integrates clinical soundness, social context, and equity to provide a structured basis for evaluation. |
| Outcome: | The proposed framework integrates clinical soundness, social context, and equity, providing a structured basis for evaluation. |
Optimizing Annotation Effort Using Active Learning Strategies: A Sentiment Analysis Case Study in Persian (2020.lrec-1)
Copied to clipboard
Seyed Arad Ashrafi Asli, Behnam Sabeti, Zahra Majdabadi, Preni Golazizian, Reza Fahmi, Omid Momenzadeh
| Challenge: | Existing deep learning approaches require huge amounts of data to be trained properly. |
| Approach: | They propose to use Persian as a model to choose the samples for annotation instead of labeling the whole dataset. |
| Outcome: | The proposed models achieve the baseline performance with a significantly lower amount of labeled data. |
Generative-to-Discriminative Test-Time Adaptation via Manifold-Aware Diffusion and Bayesian Distillation (2026.findings-acl)
Copied to clipboard
Boyun Zhang, Zequn Xie, Fangming Feng, Zihan Zhang, Yongbo He, Chuxin Wang, Sihang Cai, Tao Jin, Qifei Zhang
| Challenge: | Existing discriminative approaches suffer from "confident but wrong" failure mode, blindly adapting to OOD noise leading to error accumulation. |
| Approach: | They propose a TTA framework that harmonizes the robustness of generative diffusion models with the efficiency of discriminative regression networks via Bayesian Diffusion Distillation (BDD). |
| Outcome: | The proposed framework reduces MAE from 0.6872 to 0.5673 and boosts binary accuracy by 5.81 percentage points (reaching 57.33%) it also reduces the MAE of the MOSI to SIMS shift and achieves an 11.18-point gain over the baseline. |
Controlled Generation with Prompt Insertion for Natural Language Explanations in Grammatical Error Correction (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing studies present tokens, examples, and hints for corrections, but do not directly explain the reasons in natural language. |
| Approach: | They propose a method called controlled generation with Prompt Insertion that uses Large Language Models to explain the reasons for corrections in natural language. |
| Outcome: | The proposed method can explain the reasons for corrections in natural language by guiding the LLMs to generate explanations for all correction points. |
FaVIQ: FAct Verification from Information-seeking Questions (2022.acl-long)
Copied to clipboard
| Challenge: | Existing fact verification datasets with crowdsourced claims introduce subtle biases that are difficult to control for. |
| Approach: | They construct a large-scale fact verification dataset with ambiguous questions . they use a corpus of 188k claims to construct false and true claims . |
| Outcome: | The proposed dataset outperforms models trained on the dataset FEVER or in-domain data by up to 17% absolute. |
Analyzing Stereotypes in Generative Text Inference Tasks (2021.findings-acl)
Copied to clipboard
| Challenge: | Social psychology studies how social stereotypes are shared as part of cultural knowledge . |
| Approach: | They study how stereotypes manifest when potential targets are situated in neutral contexts . they collect human judgments on the presence of stereotypes in generated inferences based on annotator positionality . |
| Outcome: | The results show that the annotators' positions differ depending on the type of inferences they generate . |
Infusing Hierarchical Guidance into Prompt Tuning: A Parameter-Efficient Framework for Multi-level Implicit Discourse Relation Recognition (2023.acl-long)
Copied to clipboard
| Challenge: | Multi-level implicit discourse relation recognition (MIDRR) aims at identifying hierarchical discourse relations among arguments. |
| Approach: | They propose a prompt-based multi-level implicit discourse relation recognition framework that leverages parameter-efficient prompt tuning to drive inputted arguments to match the pre-trained space. |
| Outcome: | The proposed framework achieves comparable results on PDTB 2.0 and 3.0 using about 0.1% trainable parameters compared with baselines. |
Structurally Diverse Sampling for Sample-Efficient Training and Comprehensive Evaluation (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Existing approaches to generalize compositionally are inadequate, but there is no evidence for this. |
| Approach: | They propose a model-agnostic algorithm for subsampling instances with diverse structures from a labeled instance pool with structured outputs. |
| Outcome: | The proposed algorithm leads to comparable or better generalization than prior algorithms in 9 out of 10 dataset-split type pairs. |
RE-Matching: A Fine-Grained Semantic Matching Method for Zero-Shot Relation Extraction (2023.acl-long)
Copied to clipboard
Jun Zhao, WenYu Zhan, Xin Zhao, Qi Zhang, Tao Gui, Zhongyu Wei, Junzhe Wang, Minlong Peng, Mingming Sun
| Challenge: | Existing methods for zero-shot relation extraction lack explicit modeling of matching pattern . et al. (2018) show that our method achieves higher matching accuracy and faster inference speed . |
| Approach: | They propose a fine-grained semantic matching method tailored for zero-shot relation extraction . they decompose sentence-level similarity score into entity matching score and context matching score . |
| Outcome: | The proposed method achieves higher matching accuracy and faster inference speed than state-of-the-art methods. |
Noisy Label Regularisation for Textual Regression (2022.coling-1)
Copied to clipboard
| Challenge: | Existing methods to regularise noisy labels are ineffective in the face of noisy data. |
| Approach: | They propose a method that regularises noisy labels and prevents error propagation from the input layer. |
| Outcome: | The proposed method regularises noisy labels and improves generalisation performance over real-world human-disagreement annotations and randomly-corrupted and data-augmented labels. |
Towards Cross-Lingual Audio Abuse Detection in Low-Resource Settings with Few-Shot Learning (2025.coling-main)
Copied to clipboard
| Challenge: | Online abusive content detection, particularly in low-resource settings, remains underexplored. |
| Approach: | They propose to use pre-trained audio representations to detect abusive language in Indian languages using Few Shot Learning (FSL) . |
| Outcome: | The proposed model can be used to classify abusive language in 10 languages using the ADIMA dataset with FSL. |
Towards Explainable Temporal Reasoning in Large Language Models: A Structure-Aware Generative Framework (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on temporal reasoning models neglect the explainable reasoning processes underlying the results. |
| Approach: | They propose a structure-aware generative framework that integrates Graph structures with text for Explainable TEmporal Reasoning. |
| Outcome: | The proposed framework achieves state-of-the-art performance while also demonstrating robust generalization capabilities. |
GUARD: Glocal Uncertainty-Aware Robust Decoding for Effective and Efficient Open-Ended Text Generation (2025.findings-emnlp)
Copied to clipboard
Yuanhao Ding, Esteban Garces Arias, Meimingwei Li, Julian Rodemann, Matthias Aßenmacher, Danlu Chen, Gaojuan Fan, Christian Heumann, Chongsheng Zhang
| Challenge: | GUARD is a self-adaptive decoding method that balances coherence with diversity in open-ended text generation. |
| Approach: | They propose a self-adaptive decoding method that balances coherence and diversity . they combine global entropy estimates with local entropic deviations to integrate uncertainty . |
| Outcome: | GUARD achieves a good balance between diversity and coherence while exhibiting significant improvements in generation speed. |
COBRA Frames: Contextual Reasoning about Effects and Harms of Offensive Statements (2023.findings-acl)
Copied to clipboard
Xuhui Zhou, Hao Zhu, Akhila Yerukola, Thomas Davidson, Jena D. Hwang, Swabha Swayamdipta, Maarten Sap
| Challenge: | Understanding the harms and offensiveness of statements requires reasoning about the social and situational context in which they are made. |
| Approach: | They propose a context-aware formalism for explaining the intents, reactions, and harms of offensive statements grounded in their social and situational contexts. |
| Outcome: | The proposed framework is the first context-aware formalism for explaining the intents, reactions, and harms of offensive statements grounded in their social and situational context. |
Joint Modelling of Emotion and Abusive Language Detection (2020.acl-main)
Copied to clipboard
| Challenge: | Existing methods for abuse detection focus on linguistic properties of comments and online communities of users, disregarding the emotional state of the users and how this might affect their language. |
| Approach: | They propose to combine emotion and abusive language detection to create a multi-task learning framework that allows one task to inform the other. |
| Outcome: | The proposed model improves on the previous models, incorporating affective features into the learning framework. |
“Women do not have heart attacks!” Gender Biases in Automatically Generated Clinical Cases in French (2025.findings-naacl)
Copied to clipboard
| Challenge: | Healthcare professionals are increasingly including Language Models (LMs) in clinical practice. |
| Approach: | They propose to use LMs to generate clinical cases in french and an automatic linguistic gender detection tool to measure gender biases. |
| Outcome: | The proposed model over-generates cases describing male patients, creating synthetic corpora that are not consistent with documented prevalence for these disorders. |
XAutoLM: Efficient Fine-Tuning of Language Models via Meta-Learning and AutoML (2025.emnlp-main)
Copied to clipboard
Ernesto Luis Estevanell Valladares, Suilan Estevez-Velarde, Yoan Gutierrez, Andrés Montoyo, Ruslan Mitkov
| Challenge: | XAutoLM is a meta-learning-augmented framework that can be used to optimize discriminative and generative LM fine-tuning pipelines. |
| Approach: | They propose a meta-learning-augmented AutoML framework that reuses past experiences to optimize discriminative and generative LM fine-tuning pipelines efficiently. |
| Outcome: | XAutoLM surpasses zero-shot optimizer’s peak F1 on five of six tasks, reduces mean evaluation time of pipelines by up to 4.5x, and uncovers 50% more pipelines above zero- shot Pareto front. |
Adversarial Concept Erasure in Kernel Space (2022.emnlp-main)
Copied to clipboard
| Challenge: | Large neural networks in NLP produce real-valued representations that encode the bit of human language that they were trained on. |
| Approach: | They propose a kernelization of the recently-proposed linear concept-removal objective and propose to remove linear subspaces from the representation space. |
| Outcome: | The proposed kernelization protects against the ability of nonlinear adversaries to recover the concept. |
Hi-GEC: Hindi Grammar Error Correction in Low Resource Scenario (2025.coling-main)
Copied to clipboard
| Challenge: | Automated Grammatical Error Correction (GEC) is a scarcely explored low-resource language . a recent study focused on English, but it focused on Hindi, which presents unique challenges due to its complex syntax and intricate morphology. |
| Approach: | They propose to use a human-edited dataset to generate Hindi GEC data . they also investigate round trip translation using diverse languages for the technique . |
| Outcome: | The proposed method outperforms other methods in Hindi, showing that it is highly efficient. |
A Mousetrap: Fooling Large Reasoning Models for Jailbreak with Chain of Iterative Chaos (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Reasoning Models (LRMs) have advanced beyond traditional Large Language Models, yet they pose heightened safety risks. |
| Approach: | They propose a first jailbreak attack targeting Large Reasoning Models . they exploit a Chaos Machine component to transform attack prompts with diverse one-to-one mappings based on the reasoning chain . |
| Outcome: | The proposed attack exploits the unique vulnerabilities of LRMs by integrating a Chaos Machine. success rates of the mousetrap attack are as high as 96%, 86% and 98% respectively. |
REInstruct: Building Instruction Data from Unlabeled Corpus (2024.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for annotating instruction data are expensive and difficult to scale. |
| Approach: | They propose a method to automatically build instruction data from an unlabeled corpus without heavy reliance on proprietary LLMs and human annotation. |
| Outcome: | The proposed method outperforms existing methods on AlpacaEval leaderboard and other open-source methods. |
Enhancing Grammatical Error Correction Systems with Explanations (2023.acl-long)
Copied to clipboard
| Challenge: | To help language learners better understand why the GEC system makes a correction, the causes of errors and the corresponding error types are two key factors. |
| Approach: | They propose to annotate large dataset with evidence words and grammatical error types to help language learners better understand corrections. |
| Outcome: | The proposed model can be validated by human evaluation and can be used to help second-language learners decide whether to accept a correction suggestion and understand the associated grammar rule. |
Topology Imbalance and Relation Inauthenticity Aware Hierarchical Graph Attention Networks for Fake News Detection (2022.coling-1)
Copied to clipboard
| Challenge: | Existing methods to detect fake news focus on mining lexical and syntactic features. |
| Approach: | They propose a topology imbalance and Relation inauthenticity aware Hierarchical Graph Attention Networks to identify fake news on social media. |
| Outcome: | The proposed method outperforms state-of-the-art methods on real-world datasets. |
Semi-Supervised Reward Modeling via Iterative Self-Training (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Reward models capture values and preferences of humans and are used in Reinforcement Learning with Human Feedback (RLHF) Traditionally, training large language models relies on extensive human-annotated preference data, which poses significant challenges in terms of scalability and cost. |
| Approach: | They propose a method that enhances RM training using unlabeled data. |
| Outcome: | The proposed approach improves reward models without incurring additional labeling costs on unlabeled datasets. |
Do We Really Need Fully Unsupervised Cross-Lingual Embeddings? (D19-1)
Copied to clipboard
| Challenge: | a series of bilingual lexicon induction experiments with 15 diverse languages (210 language pairs) show that fully unsupervised CLWE methods fail for a large number of language pairs. |
| Approach: | They propose to use fully unsupervised approaches to project monolingual embeddings into a shared cross-lingual space without any cross-linguistic signal. |
| Outcome: | The proposed methods fail for a large number of language pairs, but never surpass weakly supervised methods. |
A Fine-Grained Taxonomy of Replies to Hate Speech (2023.emnlp-main)
Copied to clipboard
| Challenge: | a new corpus of responses to hate speech is developed to counter hate speech . authors work with real, user-generated hate speech and all the replies it elicits . counterspeech refers to a "direct response that counters hate speech" |
| Approach: | They propose a taxonomy of responses to hate speech and a new corpus to analyze responses . they find that responses to user-generated hate speech are more effective than replies generated by a third party . |
| Outcome: | The proposed taxonomy of responses to hate speech and a new corpus provide insights into content real users reply with and which replies are empirically most effective. |
LLM Distillation for Efficient Few-Shot Multiple Choice Question Answering (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Large Language Models excel at few-shot learning but their direct application in real-world scenarios is often hindered by their high computational cost. |
| Approach: | They propose a framework that uses Large Language Models for data generation and scoring to improve encoder model performance. |
| Outcome: | The proposed approach improves accuracy from 28.9% to 39.3% on a few-shot MCQA task . |
Noise-robust Cross-modal Interactive Learning with Text2Image Mask for Multi-modal Neural Machine Translation (2022.coling-1)
Copied to clipboard
| Challenge: | Existing studies on multi-modal neural machine translation focus on visual information, but text and image may not match exactly, and visual noise is often ignored. |
| Approach: | They propose a noise-robust multi-modal interactive fusion approach with cross-modal relation-aware mask mechanism for MNMT. |
| Outcome: | The proposed model achieves state-of-the-art scores in all En-De, En-Fr and En-Cs translation tasks. |
From Zero to Hero: Cold-Start Anomaly Detection (2024.findings-acl)
Copied to clipboard
| Challenge: | Existing anomaly detection methods require previous observations to be effective . contaminated observations are often not observed, making them ineffective . |
| Approach: | They propose a method that adapts a zero-shot anomaly detector to contaminated observations . they propose an evaluation suite consisting of evaluation protocols and metrics . |
| Outcome: | The proposed method adapts the zero-shot anomaly detector to contaminated observations. |
SelfPrompt: Autonomously Evaluating LLM Robustness via Domain-Constrained Knowledge Guidelines and Refined Adversarial Prompts (2025.coling-main)
Copied to clipboard
| Challenge: | Existing frameworks for evaluating robustness of large language models rely on standardized benchmarks that can escalate costs and limit evaluations across domains. |
| Approach: | They propose a framework to evaluate the robustness of large language models using adversarial prompts and domain-constrained knowledge guidelines. |
| Outcome: | The proposed framework reduces dependency on conventional benchmarks and provides efficient evaluations in constrained domains. |
Unsupervised Sentence Representation Learning with Syntactically Aligned Negative Samples (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing approaches to sentence representation learning often encounter semantic inconsistencies and feature suppression. |
| Approach: | They propose a method for generating syntactically aligned negative (SAN) samples using a semantic importance-aware Masked Language Model (MLM) approach. |
| Outcome: | The proposed method produces negative samples with substantial textual overlap with the original sentences while conveying different meanings. |
Hallucination Diversity-Aware Active Learning for Text Summarization (2024.naacl-long)
Copied to clipboard
| Challenge: | Existing methods for alleviating hallucinations require costly human annotations . Existing approaches focus on a specific type of hallucinism, which limits their effectiveness . |
| Approach: | They propose a method to detect hallucinations from errors in semantic frame, discourse and content verifiability in LLM summarization using HAllucination Diversity-Aware Sampling. |
| Outcome: | The proposed framework reduces the need for costly human annotations to correct hallucinations in LLM outputs. |
Does Corpus Quality Really Matter for Low-Resource Languages? (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing work on multilingual pre-training has relied on automatically filtered versions of CommonCrawl. |
| Approach: | They propose to use tailored crawling to identify and scrape websites with high-quality content to improve representation learning in Basque. |
| Outcome: | The proposed corpus, called EusCrawl, has a much higher quality according to native annotators than the Basque portion of popular multilingual corpora like CC100 and mC4. |
Stubborn Lexical Bias in Data and Models (2023.findings-acl)
Copied to clipboard
| Challenge: | Recent work has focused on spurious correlations between features and labels in training data . but, we find strong evidence of corresponding bias in the trained models . |
| Approach: | They propose a method to reduce spurious correlations in training data by reweighting it using a large pool of extracted features. |
| Outcome: | The proposed method reduces spurious correlations in training data, but still finds strong evidence of bias in trained models. |
Explainable Depression Detection in Clinical Interviews with Personalized Retrieval-Augmented Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing systems rely on black-box neural networks, which lack interpretability, which is crucial in mental health contexts. |
| Approach: | They propose a Retrieval-augmented generation framework for Explainable depression detection that retrieves evidence from clinical interview transcripts, providing explanations for predictions. |
| Outcome: | The proposed framework retrieves evidence from clinical interview transcripts, providing explanations for predictions. |
Efficient (Soft) Q-Learning for Text Generation with Limited Good Data (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Maximum likelihood estimation (MLE) is the predominant method for training text generation models. |
| Approach: | They propose a new RL formulation for text generation from the soft Q-learning perspective using path consistency learning to combine the best of on-/off-policy updates and learn effectively from sparse reward. |
| Outcome: | The proposed approach outperforms MLE and previous RL methods in a wide range of tasks. |
CoBA: Counterbias Text Augmentation for Mitigating Various Spurious Correlations via Semantic Triples (2025.emnlp-main)
Copied to clipboard
| Challenge: | Spurious correlations are patterns that appear in datasets but do not represent genuine relationships. |
| Approach: | They propose a more general form of counterfactual data augmentation that tackles multiple biases . they propose 'CoBA' that decomposes text into subject-predicate-object triples and modifies them to disrupt spurious correlations. |
| Outcome: | The proposed framework reduces biases and strengthens out-of-distribution resilience. |
Enhancing Effectiveness and Robustness in a Low-Resource Regime via Decision-Boundary-aware Data Augmentation (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing methods to augment textual data are limited due to the discrete characteristics of the textual dataset. |
| Approach: | They propose a decision-boundary-aware data augmentation strategy to enhance robustness using pretrained language models by shifting latent features closer to the decision boundary and reconstruction to generate an ambiguous version with a soft label. |
| Outcome: | The proposed method performs better than existing methods and is extensible with curriculum data augmentation. |
PrivacyRestore: Privacy-Preserving Inference in Large Language Models via Privacy Removal and Restoration (2025.acl-long)
Copied to clipboard
| Challenge: | Existing privacy protection methods for large language models suffer from performance degradation or large inference time overhead. |
| Approach: | They propose a plug-and-play method to protect the privacy of user inputs during LLM inference . they use offline restoration vectors to train restoration vector for each privacy span type . |
| Outcome: | The proposed method can prevent the linear growth of the privacy budget. |
“That Is a Suspicious Reaction!”: Interpreting Logits Variation to Detect NLP Adversarial Attacks (2022.acl-long)
Copied to clipboard
| Challenge: | Existing methods to detect adversarial text inputs are limited in performance and are not detectable via spell checkers. |
| Approach: | They propose a model-agnostic detector of adversarial text examples that detects patterns in the logits of the target classifier when perturbing the input text. |
| Outcome: | The proposed detector improves the state-of-the-art performance in recognizing adversarial inputs and exhibits strong generalization capabilities across different NLP models, datasets, and word-level attacks. |
Powerformer: Efficient and High-Accuracy Privacy-Preserving Language Model with Homomorphic Encryption (2025.acl-long)
Copied to clipboard
| Challenge: | a new privacy-preserving language model, Powerformer, is designed to reduce computation overhead while maintaining model performance. |
| Approach: | They propose an efficient homomorphic encryption-based privacy-preserving language model . it incorporates three key techniques to optimize encrypted computations . |
| Outcome: | The proposed model achieves 45% reduction in computation time compared to state-of-the-art models . authors say the model preserves data privacy and AI capabilities in MLaaS environments . |
Improving Paraphrase Detection with the Adversarial Paraphrasing Task (2021.acl-long)
Copied to clipboard
| Challenge: | a new adversarial method of paraphrase identification is being used to identify paraphrases based on word overlap and syntax . authors propose a dataset that generates semantically equivalent but lexically and syntactically disparate paraphrase pairs . |
| Approach: | They propose an adversarial method for paraphrase identification that uses word overlap and syntax to identify paraphrases. |
| Outcome: | The proposed method improves paraphrase detection accuracy and speed of generation of datasets. |
NLKI: A Lightweight Natural Language Knowledge Integration Framework for Improving Small VLMs in Commonsense VQA Tasks (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Small vision-language models lag behind their larger generative counterparts due to lack of knowledge. |
| Approach: | They propose a framework that integrates commonsense knowledge into small vision-language models . the framework retrieves natural language facts and prompts an LLM to craft natural language explanations . |
| Outcome: | The proposed framework retrieves natural language facts and prompts an LLM to craft natural language explanations. |
Ethical Reasoning and Moral Value Alignment of LLMs Depend on the Language We Prompt Them in (2024.lrec-main)
Copied to clipboard
| Challenge: | Ethical reasoning is a crucial skill for Large Language Models (LLMs). However, moral values are not universal, but rather influenced by language and culture. |
| Approach: | They extend the study of ethical reasoning of LLMs by (CITATION) to a multilingual setup using six languages: English, Spanish, Russian, Chinese, Hindi, and Swahili. |
| Outcome: | The proposed model is based on a multilingual setup in English, Spanish, Russian, Chinese, Hindi, and Swahili. |
Beyond Single-Event Extraction: Towards Efficient Document-Level Multi-Event Argument Extraction (2024.findings-acl)
Copied to clipboard
Wanlong Liu, Li Zhou, DingYi Zeng, Yichen Xiao, Shaohuan Cheng, Chen Zhang, Grandee Lee, Malu Zhang, Wenyu Chen
| Challenge: | mainstream event argument extraction methods process each event in isolation, resulting in inefficient inference and ignoring correlations among multiple events. |
| Approach: | They propose a multi-event argument argument extraction model which extracts arguments from all events simultaneously. |
| Outcome: | The proposed model performs better on four public datasets while saving time. |
Zero-Shot Classification by Logical Reasoning on Natural Language Explanations (2023.findings-acl)
Copied to clipboard
| Challenge: | Experimental results show that CLORE is superior to baselines on zero-shot classification tasks. |
| Approach: | They propose a framework for classification by logically parsing and reasoning on natural language explanations. |
| Outcome: | The proposed framework outperforms baselines on zero-shot classification tasks. |
MultiFiT: Efficient Multi-lingual Language Model Fine-tuning (D19-1)
Copied to clipboard
| Challenge: | Pretrained language models require unlabelled data for training, while cross-lingual models underperform on low-resource languages. |
| Approach: | They propose a multi-lingual language model fine-tuning to train and fine- tune language models efficiently in their own language. |
| Outcome: | The proposed method outperforms existing models on two widely used datasets on cross-lingual classification tasks. |
Where Fact Ends and Fairness Begins: Redefining AI Bias Evaluation through Cognitive Biases (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing benchmarks conflate factual correctness and normative fairness . a model may generate responses that are factually accurate but socially unfair . |
| Approach: | They propose a benchmark to examine the boundary between fact and fair . they draw on representativeness bias, attribution bias and ingroup–outgroup bias to explain why models often misalign fact and faireness. |
| Outcome: | The proposed model is based on ten frontier models and is available on github . it is compared with a standard model that generates people of color in Nazi-era uniforms . |
Continual Learning with Semi-supervised Contrastive Distillation for Incremental Neural Machine Translation (2024.acl-long)
Copied to clipboard
| Challenge: | Multi-domain learning is a good solution for solving domain tasks but it requires retraining when adding a new domain. |
| Approach: | They propose to exploit unlabeled data from the same distributions of the older domains to avoid catastrophic forgetting. |
| Outcome: | The proposed framework exploits unlabeled data from the same distributions of the older domains to avoid catastrophic forgetting. |
The Tug of War Within: Mitigating the Fairness-Privacy Conflicts in Large Language Models (2025.acl-long)
Copied to clipboard
| Challenge: | Existing methods to enhance an LLM's privacy awareness with thousands of samples decrease its fairness awareness. |
| Approach: | They propose a training-free method to Suppress the Privacy and faIrness coupled Neurons (SPIN) which theoretically and empirically decreases the mutual information between fairness and privacy awareness. |
| Outcome: | The proposed method reduces the mutual information between fairness and privacy awareness without compromising general capabilities. |
BAM! Born-Again Multi-Task Networks for Natural Language Understanding (P19-1)
Copied to clipboard
| Challenge: | Existing methods to train multi-task neural networks outperform or even match their single-task counterparts are difficult to implement. |
| Approach: | They propose a method that uses knowledge distillation to train multi-task neural networks that outperform or even match their single-task counterparts. |
| Outcome: | The proposed method outperforms or matches single-task neural networks on the GLUE benchmark. |
Beyond the Safety Bundle: Auditing the Helpful and Harmless Dataset (2025.naacl-long)
Copied to clipboard
| Challenge: | Learning from human feedback (LHF) has been used to mitigate the harms of large language models (LLMs) but the quality of this feedback and its effectiveness as a safety mitigation technique remain unclear. |
| Approach: | They audit the Helpful and Harmless (HH) dataset by Anthropic and examine how conceptualization failures and quality issues identified in the dataset can create additional harms . |
| Outcome: | The findings highlight the need for more nuanced, context-sensitive approaches to safety mitigation in large language models. |
Reliably Bounding False Positives: A Zero-Shot Machine-Generated Text Detection Framework via Multiscaled Conformal Prediction (2025.acl-long)
Copied to clipboard
| Challenge: | Existing methods focus excessively on detection accuracy, neglecting the societal risks posed by high false positive rates (FPRs). |
| Approach: | They propose a Conformal Prediction framework that constrains the upper bound of false positive rates and introduces a real-time detection framework. |
| Outcome: | The proposed framework reduces false positive rates and improves detection performance. |
Watch Out Your Industrial Copilots: Stealthy Backdoor Attack Against LLM-Based PLC Code Generation (2026.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are being used to generate PLC code from natural language. |
| Approach: | They propose a stealthy backdoor attack framework targeting LLM-based PLC code generation . they incorporate six malicious logic injection patterns and a pipeline to refine stealthiness . |
| Outcome: | The proposed framework achieves 82.92% success rate while remaining stealthy . it bypasses quality validation and is difficult to detect . |
Beneath the Surface: Unveiling Harmful Memes with Multimodal Reasoning Distilled from Large Language Models (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for harmful meme detection ignore in-depth cognition of meme text and image . authors propose a framework for learning reasonable thoughts from LLMs for better multimodal fusion . |
| Approach: | They propose to use large language models to learn reasonable thoughts from LLMs for better multimodal fusion and lightweight fine-tuning. |
| Outcome: | The proposed approach achieves superior performance than state-of-the-art methods on the harmful meme detection task. |
WordNet Is All You Need: A Surprisingly Effective Unsupervised Method for Graded Lexical Entailment (2023.findings-emnlp)
Copied to clipboard
| Challenge: | a simple unsupervised method for predicting graded lexical entailment in English relies on WordNet . despite its simplicity, our method outperforms all previous methods using WordNet as weak supervision. |
| Approach: | They propose an unsupervised method which relies exclusively on WordNet for predicting graded lexical entailment in English. |
| Outcome: | The proposed method outperforms existing methods on the largest GLE dataset using WordNet. |
Outlier-Safe Pre-Training for Robust 4-Bit Quantization of Large Language Models (2025.acl-long)
Copied to clipboard
| Challenge: | Quantization is a practical solution for deploying Large Language Models in resource-constrained environments. |
| Approach: | They propose an outlier-safe pre-training approach that prevents outlier formation . they validate a 1.4B-parameter model on 1 trillion tokens with no outliers . |
| Outcome: | The proposed model achieves a 35.7 average score on 1 trillion tokens with 2% training overhead. |
ChunkAttention: Efficient Self-Attention with Prefix-Aware KV Cache and Two-Phase Partition (2024.acl-long)
Copied to clipboard
| Challenge: | Experiments show that ChunkAttention can speed up the self-attention kernel by 3.2-4.8 compared to the start-of-the-art implementation. |
| Approach: | They propose a prefix-aware self-attention module that can detect matching prompt prefixes across multiple requests and share their key/value tensors in memory at runtime. |
| Outcome: | The proposed module can speed up the self-attention kernel by 3.2-4.8 compared to the start-of-the-art implementation, with the length of the system prompt ranging from 1024 to 4096. |
Cross-Domain Fake News Detection based on Dual-Granularity Adversarial Training (2025.coling-main)
Copied to clipboard
| Challenge: | Existing approaches to detect fake news in unseen domains are limited by domain-specific training. |
| Approach: | They propose a cross-domain fake news detection method based on adversarial training . they use a document-level and entity-level model to generate domain-independent representations . |
| Outcome: | The proposed method can detect fake news in unseen domains with the help of pre-trained language models. |
Do What? Teaching Vision-Language-Action Models to Reject the Impossible (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Recent studies show that VLAs can recognize, interpret, and respond to false-premise instructions. |
| Approach: | They propose a framework that detects when an instruction cannot be executed due to a false premise and engages in language-based clarification or correction. |
| Outcome: | The proposed framework detects when an instruction cannot be executed due to a false premise and engages in language-based clarification or correction. |
AS-ES Learning: Towards efficient CoT learning in small models (2024.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to induce Chain-of-Thought (CoT) in LLMs are limited and do not consider the importance of efficiently utilizing existing CoT data. |
| Approach: | They propose a new training paradigm which exploits the inherent information in CoT for iterative generation. |
| Outcome: | The proposed training paradigm surpasses direct seq2seq training on CoT-extensive tasks without data augmentation or altering the model itself. |
Finding Spoken Identifications: Using GPT-4 Annotation for an Efficient and Fast Dataset Creation Pipeline (2024.lrec-main)
Copied to clipboard
Maliha Jahan, Helin Wang, Thomas Thebaud, Yinglun Sun, Giang Ha Le, Zsuzsanna Fagyal, Odette Scharenborg, Mark Hasegawa-Johnson, Laureano Moro Velazquez, Najim Dehak
| Challenge: | Existing datasets that are limited to a few dialects, ethnicities, and age groups are not annotated considering these factors. |
| Approach: | They propose a semi-automated dataset creation pipeline that leverages large language models to perform two complex annotation tasks using human annotations as ground truths. |
| Outcome: | The proposed pipeline reduces time required for the filtering and tagging tasks while losing no important information. |
Multi-source Meta Transfer for Low Resource Multiple-Choice Question Answering (2020.acl-main)
Copied to clipboard
| Challenge: | Existing MCQA datasets are small in size, which increases difficulty of model learning and generalization. |
| Approach: | They propose a multi-source meta transfer framework for low-resource multiple-choice question answering . they extend meta learning by incorporating multiple training sources to learn a generalized feature representation across domains . |
| Outcome: | The proposed framework is independent of backbone language models and can bridge the distribution gap between training sources and target. |
Cross-Lingual Cross-Target Stance Detection with Dual Knowledge Distillation Framework (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing studies on stance detection were conducted mainly in English due to the low-resource problem in most non-English languages. |
| Approach: | They propose to use a cross-lingual teacher and a teacher to transfer knowledge from source to target language to bridge the discrepancy between languages. |
| Outcome: | The proposed framework bridges the discrepancy between languages and generalizes the knowledge to unseen targets in target language. |
LightThinker: Thinking Step-by-Step Compression (2025.emnlp-main)
Copied to clipboard
Jintian Zhang, Yuqi Zhu, Mengshu Sun, Yujie Luo, Shuofei Qiao, Lun Du, Da Zheng, Huajun Chen, Ningyu Zhang
| Challenge: | Recent advances in Large Language Models have demonstrated their remarkable capabilities in complex reasoning tasks, but their efficiency is hindered by the substantial memory and computational costs associated with generating lengthy tokens. |
| Approach: | They propose a method that dynamically compresses verbose thought steps into compact representations and discards original reasoning chains. |
| Outcome: | The proposed method reduces peak memory usage and inference time while maintaining competitive accuracy. |
A Shared Task of a New, Collaborative Type to Foster Reproducibility: A First Exercise in the Area of Language Science and Technology with REPROLANG2020 (2020.lrec-1)
Copied to clipboard
António Branco, Nicoletta Calzolari, Piek Vossen, Gertjan Van Noord, Dieter van Uytvanck, João Silva, Luís Gomes, André Moreira, Willem Elbers
| Challenge: | Scientific knowledge is grounded on falsifiable predictions and therefore its credibility and raison d'être rely on the possibility of repeating experiments and getting similar results as originally obtained and reported. |
| Approach: | They propose a collaborative task which is collaborative rather than competitive and supports reproduction of research results. |
| Outcome: | The proposed task is called REPROLANG-The Shared Task on the Reproduction of Research Results in Science and Technology of Natural Language Processing (LREC2020). |
MuTIS: Enhancing Reasoning Efficiency through Multi Turn Intervention Sampling in Reinforcement Learning (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for training large reasoning models with long chain-of-thought (CoT) are limited by the number of parameters and the complexity of the model. |
| Approach: | They propose a framework that leverages multi-turn interventions to produce concise reasoning chains and demonstrates strong scalability. |
| Outcome: | The proposed framework breaks the accuracy-efficiency trade-off by producing concise reasoning chains and demonstrating strong scalability on 7B models. |
To Lie or Not to Lie? Investigating The Biased Spread of Global Lies by LLMs (2026.acl-long)
Copied to clipboard
Zohaib Khan, Mustafa Dogan, Ifeoma Okoh, Pouya Sadeghi, Siddhartha Shrestha, Sergius Justus Chesami Nyah, Mahmoud O. Mokhiamar, Michael J Ryan, Tarek Naous
| Challenge: | Misinformation is on the rise, and the strong writing capabilities of LLMs lower the barrier for malicious actors to produce and disseminate false information. |
| Approach: | They introduce a multilingual parallel dataset of 440 misinformation generation prompt templates and 6,867 entities, spanning 8 languages and 195 countries. |
| Outcome: | The proposed model reduces misinformation generation across languages and countries . it also reduces the risk of misinformation being spread across countries based on the model's performance . |
Compositional Generalisation for Explainable Hate Speech Detection (2025.emnlp-main)
Copied to clipboard
| Challenge: | Hate speech detection models struggle to generalise beyond their training data . dataset biases and the use of sentence-level labels fail to teach the underlying structure of hate speech. |
| Approach: | They propose to use a dataset to train models with fine-grained, span-level annotations . they find that combinations of expressions that deviate from those seen during training are difficult to detect . |
| Outcome: | The proposed model can generalise to a dataset with equal frequency across all contexts while achieving state-of-the-art performance on the human-sourced PLEAD. |
Guided by Gut: Efficient Test-Time Scaling with Reinforced Intrinsic Confidence (2026.acl-long)
Copied to clipboard
| Challenge: | Guided by Gut (GG) is an efficient self-guided TTS framework for Large Language Models (LLMs) that performs step-by-step reasoning at a low cost without any reward models or verifiers. |
| Approach: | They propose a self-guided TTS framework that enables LLMs to perform step-by-step reasoning at a low cost without any reward models or verifiers. |
| Outcome: | Empirical evaluations show that GG performs better than TTS with PRMs while reducing GPU memory usage by up to 10. |
Faithful Knowledge Graph Explanations in Commonsense Question Answering (2022.emnlp-main)
Copied to clipboard
| Challenge: | Knowledge graphs are used to express explanations for the model's answer choice. |
| Approach: | They propose to use knowledge graphs to encode facts separately from the question and combine them to select an answer. |
| Outcome: | The proposed architectures can be used to express the facts used to answer a question in a graph-based explanation, but they will not include reasoning done by the transformer encoding the question, and will be incomplete. |
Demonstrations Are All You Need: Advancing Offensive Content Paraphrasing using In-Context Learning (2024.findings-acl)
Copied to clipboard
| Challenge: | Paraphrasing of offensive content is a better alternative to content removal, but supervised methods often retain a large portion of the offensiveness of the original content. |
| Approach: | They propose to use In-Context Learning (ICL) to generate usable offensive paraphrases by using large language models. |
| Outcome: | The proposed framework is better than supervised methods on human evaluation and lower toxicity by 76%. |
Monitoring Decoding: Mitigating Hallucination via Evaluating the Factuality of Partial Response during Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to mitiga hallucinations rely on sampling multiple full-length generations, which introduces significant response latency and becomes ineffective when the model consistently produces hallucines. |
| Approach: | They propose a framework that dynamically monitors the generation process and selectively applies in-process interventions to revise hallucination-prone tokens. |
| Outcome: | The proposed framework outperforms self-consistency-based approaches in both effectiveness and efficiency, achieving higher factual accuracy while significantly reducing computational overhead. |
StereoMap: Quantifying the Awareness of Human-like Stereotypes in Large Language Models (2023.emnlp-main)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have been observed to encode harmful associations present in the training data. |
| Approach: | They propose a framework to map LLMs' perceptions of how demographic groups have been viewed by society using the dimensions of Warmth and Competence. |
| Outcome: | The proposed framework maps LLMs’ perceptions of social groups using the dimensions of Warmth and Competence. |
ReCode: Robustness Evaluation of Code Generation Models (2023.acl-long)
Copied to clipboard
Shiqi Wang, Zheng Li, Haifeng Qian, Chenghao Yang, Zijian Wang, Mingyue Shang, Varun Kumar, Samson Tan, Baishakhi Ray, Parminder Bhatia, Ramesh Nallapati, Murali Krishna Ramanathan, Dan Roth, Bing Xiang
| Challenge: | Existing work on robustness in text or code tasks has focused on classification, while robustness for code generation tasks is an uncharted area. |
| Approach: | They propose a robustness evaluation benchmark for code generation models that customizes over 30 transformations specifically for code on docstrings, function and variable names, code syntax, and code format. |
| Outcome: | The proposed model performs better on human annotators and on SOTA models with human annnotators. |
SecFormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via SMPC (2024.findings-acl)
Copied to clipboard
| Challenge: | a growing number of cloud-based inference services are relying on SMPC to protect data privacy. |
| Approach: | They propose a framework for Privacy-Preserving Inference for Transformer models that eliminates exponential and maximum operations in PPI without sacrificing model performance. |
| Outcome: | The proposed framework outperforms MPCFormer in terms of performance and efficiency . it is 3.57 and 3.58 times faster than PUMA for BERTBASE and BERTLARGE . |
“Are Your Explanations Reliable?” Investigating the Stability of LIME in Explaining Text Classifiers by Marrying XAI and Adversarial Attack (2023.emnlp-main)
Copied to clipboard
| Challenge: | Extensive experiments on text datasets demonstrate that XAIFooler significantly outperforms all baselines by large margins in its ability to manipulate LIME’s explanations with high semantic preservability. |
| Approach: | They propose to use LIME to establish a baseline and then propose an algorithm to perturb text inputs and manipulate explanations. |
| Outcome: | The proposed algorithm outperforms baselines on text datasets and achieves high semantic preservability. |
Evaluating distillation methods for data-efficient syntax learning (2025.findings-emnlp)
Copied to clipboard
| Challenge: | knowledge distillation (KD) targeting attention should selectively accelerate syntax acquisition, a study finds . logit-based KD dramatically improves data-efficiency, attention-based one provides minimal benefit even for syntactic tasks. |
| Approach: | a study predicts that knowledge distillation targeting attention should selectively accelerate syntax acquisition . a systolic analysis of student models compared to logit-based knowledge distillations . |
| Outcome: | a new study shows that knowledge distillation (KD) targeting attention accelerates syntax acquisition . the hypothesis is tested on syntactic benchmarks and perplexity. |
EXPLAIN, EDIT, GENERATE: Rationale-Sensitive Counterfactual Data Augmentation for Multi-hop Fact Verification (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods to augment training data with counterfactuals fail to handle multi-hop fact verification due to their incapability to preserve complex logical relationships. |
| Approach: | They propose to augment training data with counterfactuals that alter causal features of the original data by preserving logical relationships. |
| Outcome: | The proposed method outperforms the baselines and can generate linguistically diverse counterfactuals without disrupting their logical relationships. |
Generalizable Cross-Lingual Cognitive Distortion Detection with Standardized Annotations and Multi-Task Learning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on cognitive distortion have limited generalizability and performance of models in large-scale and cross-linguistic contexts. |
| Approach: | They propose a multi-task learning model based on teacher student architecture solution which improves generalization performance. |
| Outcome: | The proposed model improves generalizability and interpretability of the proposed model. |
Shortcuts Arising from Contrast: Towards Effective and Lightweight Clean-Label Attacks in Prompt-Based Learning (2024.emnlp-main)
Copied to clipboard
| Challenge: | Prompt-based learning paradigms are vulnerable to backdoor attacks, requiring false activations and false data augmentation. |
| Approach: | They propose a method that uses triggers to create stronger shortcuts by leveraging activation values and data selection strategies to create the shortcuts. |
| Outcome: | The proposed method is based on the concept that a backdoor acts as a shortcut and can achieve high effectiveness and stealthiness at low poisoning rates. |
Dual-Stage Multi-Task Syntax-Oriented Pre-Training for Syntactically Controlled Paraphrase Generation (2024.findings-acl)
Copied to clipboard
| Challenge: | Syntactically controlled paraphrase generation (SCPG) aims to generate sentences with syntactic structures resembling given exemplars. |
| Approach: | They propose a dual-stage multi-task pre-training scheme that uses a series of structure-oriented and syntax-oriented tasks to generate sentences with syntactic structures resembling given exemplars. |
| Outcome: | The proposed method outperforms existing methods on all possible variants of SCPG tasks and significantly outperformed the popular T5 model. |
AGD: Adversarial Game Defense Against Jailbreak Attacks in Large Language Models (2025.acl-long)
Copied to clipboard
Shilong Pan, Zhiliang Tian, Zhen Huang, Wanlong Yu, Zhihua Wen, Xinwang Liu, Kai Lu, Minlie Huang, Dongsheng Li
| Challenge: | Existing defenses, including post-training alignment and prompt engineering, struggle with adaptability to out-of-distribution (OOD) attacks. |
| Approach: | They propose an adversarial game-based defense method that dynamically adjusts LLMs’ internal representations to achieve a balanced trade-off between helpfulness and harmlessness. |
| Outcome: | The proposed method improves LLMs’ safety over all baselines. |
From Tokens to Steps: Verification-Aware Speculative Decoding for Efficient Multi-Step Reasoning (2026.findings-acl)
Copied to clipboard
| Challenge: | Speculative decoding (SD) allows a lightweight draft model to propose outputs that a stronger target model verifies. |
| Approach: | They propose a verification-aware speculative decoding framework that performs step-level verification using only model-internal signals. |
| Outcome: | Experiments show that SpecGuard outperforms both SD and reward-guided SD in accuracy and reliability tests. |
SMoP: Towards Efficient and Effective Prompt Tuning with Sparse Mixture-of-Prompts (2023.emnlp-main)
Copied to clipboard
| Challenge: | Prompt tuning has emerged as a successful parameter-efficient alternative to the full fine-tuning of language models. |
| Approach: | They propose a prompt tuning method that utilizes short soft prompts for efficient training and inference while maintaining performance gains typically induced by longer soft prompt. |
| Outcome: | The proposed method outperforms baseline methods while preserving memory usage. |
Ethical Reasoning over Moral Alignment: A Case and Framework for In-Context Ethical Policies in LLMs (2023.findings-emnlp)
Copied to clipboard
| Challenge: | a paper by a team of researchers proposes that large language models should be morally aligned to ethical principles . a moral compass is a model that integrates moral dilemmas with moral principles pertaining to different foramlisms of normative ethics . |
| Approach: | They propose to infuse generic ethical reasoning capabilities into large-scale models . they argue that LLMs should take a moral stance on value pluralism . |
| Outcome: | a new ethical reasoning framework integrates moral dilemmas with moral principles . the framework is based on the results of a hypothetical case study on a large-scale model . |
MLLM-Protector: Ensuring MLLM’s Safety without Hurting Performance (2024.emnlp-main)
Copied to clipboard
Renjie Pi, Tianyang Han, Jianshu Zhang, Yueqi Xie, Rui Pan, Qing Lian, Hanze Dong, Jipeng Zhang, Tong Zhang
| Challenge: | MLLMs are deployed on limited image-text pairs, which makes them more vulnerable to catastrophic forgetting of their original abilities during safety fine-tuning. |
| Approach: | They propose a plug-and-play strategy that detects harmful visual inputs and transforms harmful ones into harmless ones. |
| Outcome: | The proposed approach mitigates the risks posed by malicious visual inputs without compromising the original performance of MLLMs. |
XC-Cache: Cross-Attending to Cached Context for Efficient LLM Inference (2024.findings-emnlp)
Copied to clipboard
Joao Monteiro, Étienne Marcotte, Pierre-Andre Noel, Valentina Zantedeschi, David Vazquez, Nicolas Chapados, Christopher Pal, Perouz Taslakian
| Challenge: | XC-Llama uses pre-trained decoder-only models to condition generation on reference text without the prompt. |
| Approach: | They propose a model that uses cross-attention to condition generation on reference text without the prompt. |
| Outcome: | The proposed models outperform prompt-based inference methods and reduce space footprint relative to standard KV caching by two orders of magnitude. |
Mapping the Minds of LLMs: A Graph-Based Analysis of Reasoning LLMs (2025.emnlp-main)
Copied to clipboard
| Challenge: | Large Reasoning Models (LRMs) often display unstable behaviors, e.g., hallucinating unsupported premises, overthinking simple tasks, and displaying higher sensitivity to prompt variations. |
| Approach: | They propose a graph-based analytical framework that clusters long, verbose CoT outputs into semantically coherent reasoning steps, then constructs directed reasoning graphs to capture contextual and logical dependencies among these steps. |
| Outcome: | The proposed framework enables quantitative evaluation of internal reasoning structure and quality beyond conventional metrics and provides practical insights for prompt engineering and cognitive analysis of LLMs. |
ACCEPT: Adaptive Codebook for Composite and Efficient Prompt Tuning (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Prompt Tuning has been a popular fine-tuning method for large-scale pretrained language models. |
| Approach: | They propose a method that allows all soft prompts to share a set of learnable codebook vectors in each subspace, with each prompt differentiated by a number of adaptive weights. |
| Outcome: | The proposed method achieves superior performance on 17 diverse natural language tasks including natural language understanding (NLU) and question answering (QA) tasks by tuning only 0.3% of parameters of the PLMs. |
Beyond Demographics: Enhancing Cultural Value Survey Simulation with Multi-Stage Personality-Driven Cognitive Reasoning (2025.emnlp-main)
Copied to clipboard
| Challenge: | Introducing **MARK**, a framework for cultural value survey simulation . based on type dynamics theory, it improves accuracy and interpretation of models . |
| Approach: | They propose a framework that integrates psychological theory into cultural value survey simulations. |
| Outcome: | The proposed framework outperforms baseline models on the World Values Survey by 10% accuracy and reduces divergence between model predictions and human preferences. |
FACTIFY3M: A benchmark for multimodal fact verification with explainability through 5W Question-Answering (2023.emnlp-main)
Copied to clipboard
Megha Chakraborty, Khushbu Pahwa, Anku Rani, Shreyas Chatterjee, Dwip Dalal, Harshit Dave, Ritvik G, Preethi Gurumurthy, Adarsh Mahor, Samahriti Mukherjee, Aditya Pakala, Ishan Paul, Janvita Reddy, Arghya Sarkar, Kinjal Sensharma, Aman Chadha, Amit Sheth, Amitava Das
| Challenge: | Disinformation can cause disruption in the share market, panic and anxiety in society, and even death during crises. |
| Approach: | a new dataset is being developed to help combat disinformation . the dataset is a multimodal fake news dataset with 5W question-answering . |
| Outcome: | FACTIFY 3M is the largest dataset and benchmark for multimodal fact verification. |
Accept or Deny? Evaluating LLM Fairness and Performance in Loan Approval across Table-to-Text Serialization Approaches (2025.findings-emnlp)
Copied to clipboard
Israel Abebe Azime, Deborah D. Kanubala, Tejumade Afonja, Mario Fritz, Isabel Valera, Dietrich Klakow, Philipp Slusallek
| Challenge: | Large Language Models (LLMs) are increasingly employed in high-stakes decision-making tasks such as loan approvals. |
| Approach: | They evaluate the performance and fairness of LLMs on serialized loan approval datasets from Ghana, Germany, and the United States. |
| Outcome: | The model’s zero-shot and in-context learning (ICL) capabilities are evaluated on loan approval datasets from Ghana, Germany, and the United States. |
K-HATERS: A Hate Speech Detection Corpus in Korean with Target-Specific Ratings (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing datasets on hate speech detection focus on overt forms of hate . however, a majority of these resources are English-centric, focusing on overtones of hate. |
| Approach: | They propose a new corpus for hate speech detection in Korean with target-specific offensiveness ratings that offer a three-point Likert scale. |
| Outcome: | The proposed corpus is the largest offensive language corpus in Korean and offers target-specific ratings on a three-point Likert scale. |
MiniKV: Pushing the Limits of 2-Bit KV Cache via Compression and System Co-Design for Efficient Long Context Inference (2025.findings-acl)
Copied to clipboard
| Challenge: | State-of-the-art 2-bit KV cache quantization methods achieve excellent results in accelerating LLM inference while retaining accuracy on long context tasks. |
| Approach: | They propose a method based on 2-bit KV cache quantization with adaptive KV policies that retain LLM accuracy with only a subset of KV states. |
| Outcome: | The proposed method outperforms state-of-the-art methods on a wide range of long context tasks while retaining accuracy. |
Multilingual Text-to-Image Generation Magnifies Gender Stereotypes (2025.acl-long)
Copied to clipboard
Felix Friedrich, Katharina Hämmerl, Patrick Schramowski, Manuel Brack, Jindřich Libovický, Alexander Fraser, Kristian Kersting
| Challenge: | Text-to-image (T2I) generation models have great results in image quality, flexibility, and text alignment, but they suffer from substantial gender bias. |
| Approach: | They propose a benchmark to study gender bias in multilingual T2I models . they use multilingual prompts to account for grammatical differences influencing gender . |
| Outcome: | The proposed benchmark shows strong gender biases and language-specific differences across models. |
The Lies Characters Tell: Utilizing Large Language Models to Normalize Adversarial Unicode Perturbations (2025.findings-acl)
Copied to clipboard
| Challenge: | Homoglyphs are visually homogeneous to Latin letters and are used to mask offensive content. |
| Approach: | They propose two methods to normalize homoglyphs by replacing non-Latin characters with a delimiter and using large language models to determine which characters should be replaced with Latin letters. |
| Outcome: | The proposed methods normalize homoglyphs by replacing non-Latin characters with a delimiter and prompting large language models to "fill in the blanks" the authors found that the proposed methods produced normalized text with an average cosine similarity score of 0.91 to the original tweets and 0.96 to the tweets using the direct method. |
Adaptive LoRA Merge with Parameter Pruning for Low-Resource Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for adapting LLMs to low-resource tasks keep LoRA parameters frozen and the low-level problem out of their scope. |
| Approach: | They propose a LoRA merge method that updates and prunes LoRA parameters through fine-tuning with minimal target task data. |
| Outcome: | The proposed method improves performance on a low-resource language generation task and improves on previous methods. |
Annotation Sensitivity: Training Data Collection Methods Affect Model Performance (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Using an annotation instrument, the design of the annotation instrument and the instructions given to annotators can impact training data. |
| Approach: | They investigate the impact of an annotation instrument on training data . they collect hate speech and offensive language annotations in a tweet corpus . |
| Outcome: | The proposed model performs better on holdout conditions than on the standard model. |
The Elephant in the Room: Exploring the Role of Neutral Words in Language Model Group-Agnostic Debiasing (2025.findings-acl)
Copied to clipboard
Xinwei Guo, Jiashi Gao, Junlei Zhou, Jiaxin Zhang, Guanhua Chen, Xiangyu Zhao, Quanying Liu, Haiyan Wu, Xin Yao, Xuetao Wei
| Challenge: | Large Language Models (LLMs) are increasingly integrated into our daily lives, raising ethical concerns, especially about perpetuating stereotypes. |
| Approach: | They propose a method that incorporates a neutral word semantics-based loss function to alleviate the deterioration of the LMS during debiasing. |
| Outcome: | The proposed method alleviates the deterioration of the Language Modeling Score (LMS) by incorporating a neutral word semantics-based loss function. |
Attention-guided Self-reflection for Zero-shot Hallucination Detection in Large Language Models (2025.emnlp-main)
Copied to clipboard
| Challenge: | Hallucination is a significant barrier to the effective application of Large Language Models (LLMs). |
| Approach: | They propose an Attention-Guided SElf-Reflection approach for hallucination detection in Large Language Models. |
| Outcome: | The proposed method significantly outperforms existing methods in zero-shot hallucination detection on four widely-used LLMs across three different halluciation benchmarks. |
CaT-Bench: Benchmarking Language Model Understanding of Causal and Temporal Dependencies in Plans (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing studies on reasoning in plans focus on classical problems, simulated environments, or restricted language such as PDDL, but real-world plans cannot be tested to test for correctness and reliability. |
| Approach: | They propose a benchmark question that tests whether a step must necessarily occur before or after another in cooking recipe plans. |
| Outcome: | The proposed question-driven evaluation shows that SOTA LLMs are underwhelming and biased towards predicting dependence more often, but the best F1 result is 0.73. |
RiOT: Efficient Prompt Refinement with Residual Optimization Tree (2025.acl-long)
Copied to clipboard
| Challenge: | Existing methods for automatic prompt optimization face two challenges: lack of diversity and semantic drift. |
| Approach: | They propose a framework for automatic prompt optimization that iteratively refines prompts through text gradients and selects the best prompt using perplexity. |
| Outcome: | The proposed framework outperforms existing prompt optimization methods and manual prompting on commonsense, mathematical, logical, temporal, and semantic reasoning benchmarks. |
On the Relationship between Skill Neurons and Robustness in Prompt Tuning (2024.lrec-main)
Copied to clipboard
| Challenge: | Prompt Tuning is a parameter-efficient finetuning method for pre-trained large language models (PLMs). |
| Approach: | They propose to use RoBERTa to fine tune pre-trained large language models by finetuning only a small set of parameters to adjust for downstream tasks. |
| Outcome: | The proposed method activates specific neurons in the transformer’s feed-forward networks that are highly predictive and selective for the given task. |
Stratified Selective Sampling for Instruction Tuning with Dedicated Scoring Strategy (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Recent work shows that post-training datasets can be substantially downsampled without noticeably deteriorating performance. |
| Approach: | They propose a method that efficiently bins data into groups and scores difficulty using specialized models. |
| Outcome: | The proposed method can be efficient and universally applied to post-training datasets. |
Does Rationale Quality Matter? Enhancing Mental Disorder Detection via Selective Reasoning Distillation (2025.findings-acl)
Copied to clipboard
| Challenge: | incorporating clinical symptom information into a model enhances domain expertise, improving its detection and interpretation performance. large language models are effective for generating explanatory rationales, but inconsistencies in relevance and domain alignment of LLM-generated rationale are challenging. |
| Approach: | They propose a framework that fine-tunes smaller language models with rationales exhibiting high domain relevance . they propose 'quality-focused' approach that selects rationale based on their alignment with clinical reasoning . |
| Outcome: | The proposed framework improves mental health detection and interpretation performance by ensuring high-quality rationales with domain relevance. |
Distinguishing fair from unfair compositional generalization tasks (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Compositional generalization benchmarks assess learning agents' ability to combine familiar concepts in novel ways. |
| Approach: | They propose to use compositional generalization benchmarks to assess learning agents' ability to combine familiar concepts in novel ways. |
| Outcome: | The proposed tasks are easy and hard, but no present-day models get any traction. |
Blind Men and the Elephant: Diverse Perspectives on Gender Stereotypes in Benchmark Datasets (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing benchmarks for measuring gender stereotypical bias in language models are inconsistencies . lack of explicit standards in data gathering can have detrimental effects on results . |
| Approach: | They propose that currently available benchmarks capture only partial facets of gender stereotypes . they apply a framework from social psychology to balance data across components of gender stereotypes based on stereotypical benchmarks. |
| Outcome: | The proposed framework improves correlation between different benchmarks by using simple balancing techniques. |
BiasWipe: Mitigating Unintended Bias in Text Classifiers through Model Interpretability (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods to mitigate unintended bias in social media platforms are re-training and adding extra parameters to the model. |
| Approach: | They propose a technique to mitigate unintended bias in language models by pruning the neuron weights responsible for univ bias. |
| Outcome: | The proposed technique achieves fairness by pruning the neuron weights responsible for unintended bias without loss of original performance. |
FastMCTS: A Simple Sampling Strategy for Data Synthesis (2025.acl-long)
Copied to clipboard
| Challenge: | Existing methods for generating multi-step reasoning data rely on rejection sampling, which generates trajectories independently and suffers from inefficiency and imbalanced sampling across problems of varying difficulty levels. |
| Approach: | They propose a data synthesis strategy inspired by Monte Carlo Tree Search . it offers step-level evaluation signals and promotes balanced sampling . |
| Outcome: | Experiments show that FastMCTS generates 30% more correct reasoning paths than rejection sampling. |
ConfSpec: Efficient Step-Level Speculative Reasoning via Confidence-Gated Verification (2026.acl-long)
Copied to clipboard
| Challenge: | Existing approaches to chain-of-thought reasoning incur high inference latency due to long generation traces. |
| Approach: | They propose a confidence-gated cascaded verification framework that reduces the trade-off between generation and verification. |
| Outcome: | The proposed framework achieves 2.24 speedups while matching target-model accuracy. |
Mechanistic Interpretability of Text-to-Image Diffusion Models via Cross-Attention Interventions (2026.findings-acl)
Copied to clipboard
| Challenge: | Text-to-image diffusion models generate high quality images through iterative denoising, but their internal mechanisms for grounding prompt semantics into visual structure remain unclear. |
| Approach: | They propose a mechanistic interpretability framework that probes how individual prompt tokens are represented and utilized during the denoising process. |
| Outcome: | The proposed framework enables module-wise and head-wise attribution of semantic changes across denoising timesteps. |
Steering Away from Refusal: A Black-box Jailbreak Method Based on First-Token Distribution (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to analyze black-box jailbreaks lack direct optimization signals to refine adversarial prompts. |
| Approach: | They propose a distribution-jailbreak attack method that selects effective jailbreak templates and iteratively optimizes adversarial suffixes by maximizing the KL divergence from the standard refusal distribution. |
| Outcome: | The proposed method achieves state-of-the-art Attack Success Rate (ASR) on all tested open-source models and delivers over 94% ASR on GPT-4.1. |
Harmful Prompt Laundering: Jailbreaking LLMs with Abductive Styles and Symbolic Encoding (2025.emnlp-main)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have demonstrated remarkable capabilities but their misuse for harmful purposes remains a concern. |
| Approach: | They propose a jailbreaking technique that exploits weaknesses in LLMs' architecture . they propose abductive framing and symbolic encoding to bypass safeguards . |
| Outcome: | The proposed technique achieves over 95% attack success rate on GPT-series models and 70% across all targets. |
SOBR: A Corpus for Stylometry, Obfuscation, and Bias on Reddit (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing corpora are limited in scope and can be used to collect data on author attributes. |
| Approach: | They propose to use subreddits, flairs, and self-reports as distant labels for author attributes (age, gender, nationality, personality, and political leaning) . |
| Outcome: | The proposed method could be used to infer author attributes from public posts despite their discreetness and anonymity . |
When Should Dense Retrievers Be Updated in Evolving Corpora? Detecting Out-of-Distribution Corpora Using GradNormIR (2025.findings-acl)
Copied to clipboard
Dayoon Ko, Jinyoung Kim, Sohyeon Kim, Jinhyuk Kim, Jaehoon Lee, Seonghak Song, Minyoung Lee, Gunhee Kim
| Challenge: | Dense retrievers encode text into embeddings to retrieve relevant documents . however, real-world corpora evolve, resulting in degraded retrieval performance . identifying when a dense retriever requires an update is critical for robust retrieval systems . |
| Approach: | They propose a task of predicting whether a corpus is out-of-distribution (OOD) relative to a dense retriever before indexing. |
| Outcome: | The proposed method detects whether a corpus is out-of-distribution (OOD) relative to a dense retriever before indexing. |
CLARO: Controlled Attribute-Driven Reasoning Optimization for Efficient Chain-of-Thought (2026.findings-acl)
Copied to clipboard
| Challenge: | Recent approaches guide reasoning length through token penalties or truncation, risking the omission of necessary steps. |
| Approach: | They propose a framework to internalize conciseness and attribute-driven reasoning optimization . they propose guiding models to embed high-quality structural attributes within a token budget . |
| Outcome: | The proposed method outperforms state-of-the-art models across benchmarks yielding accuracy gains of up to 63.6%. |
AMANDA: Agentic Medical Knowledge Augmentation for Data-Efficient Medical Visual Question Answering (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing Med-MLLMs fail when deployed in low-resource settings where abundant labeled data is unavailable. |
| Approach: | They propose a training-free agentic framework that performs medical knowledge augmentation via LLM agents. |
| Outcome: | The proposed framework performs medical knowledge augmentation via LLM agents. |
Mixed Signals: Decoding VLMs’ Reasoning and Underlying Bias in Vision-Language Conflict (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Vision-language models have demonstrated impressive performance by effectively integrating visual and textual information to solve complex tasks. |
| Approach: | They build upon existing benchmarks to create five datasets containing mismatched image-text pairs and examine how they reason over visual and textual data . |
| Outcome: | The proposed model reasoned over visual and textual data in real-world applications but not in the visual and visual descriptions. |
MARCH: Evaluating the Intersection of Ambiguity Interpretation and Multi-hop Inference (2026.findings-acl)
Copied to clipboard
Jeonghyun Park, Ingeol Baek, Seunghyun Yoon, Haeun Jang, Aparna Garimella, Akriti Jain, Nedim Lipka, Hwanhee Lee
| Challenge: | Existing benchmarks on multi-hop QA focus on single-hop and layered ambiguity, but they focus on ambiguous questions . ambiguities can arise at any stage, complicating the reasoning process . |
| Approach: | They propose a benchmark to evaluate ambiguity in multi-hop question answering . they propose MARCH, which uses 2,209 carefully annotated questions . |
| Outcome: | The proposed framework outperforms existing approaches and significantly outperfies existing frameworks. |
DIPLomA: Efficient Adaptation of Instructed LLMs to Low-Resource Languages via Post-Training Delta Merging (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Using delta-based weight merging, open-weight instruction-tuned large language models can be efficiently adapted to low-resource languages without costly large-scale post-training. |
| Approach: | They propose a delta-based instruction-preserving language adaptation strategy that decouples adaptation from post-training alignment by pretraining a foundational LLM on monolingual target-language data and then injecting instruction-following capabilities via delta-derived weight merging. |
| Outcome: | The proposed method achieves significant gains in instruction-following, linguistic proficiency, and safety on Basque, Welsh, and Swahili while preserving the original model's multilingual performance. |
Does It Run and Is That Enough? Revisiting Text-to-Chart Generation with a Multi-Agent Approach (2025.findings-emnlp)
Copied to clipboard
| Challenge: | 15% of generated scripts fail to execute, even after supervised fine-tuning and reinforcement learning. |
| Approach: | They propose a lightweight multi-agent pipeline that separates drafting, execution, repair, and judgment . the system reduces execution errors to 4.5% within three repair iterations . |
| Outcome: | The proposed pipeline reduces execution errors to 4.5% within three repair iterations while requiring significantly less compute. |
HASH-RAG: Bridging Deep Hashing with Retriever for Efficient, Fine Retrieval and Augmented Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Experimental evaluations on NQ, TriviaQA, and HotpotQA datasets demonstrate that our approach achieves a 90% reduction in retrieval time compared to conventional methods while maintaining considerate recall performance. |
| Approach: | They propose a framework that integrates deep hashing techniques with systematic optimizations to address these limitations. |
| Outcome: | The proposed framework outperforms retrieval/non-retrieval baselines by 1.4-4.3% in EM scores on NQ, TriviaQA, and HotpotQA datasets. |
Beyond Online Sampling: Bridging Offline-to-Online Alignment via Dynamic Data Transformation for LLMs (2025.emnlp-main)
Copied to clipboard
| Challenge: | Direct Preference Optimization (DPO) eliminates complex reward modeling in aligning large language models with human preferences, but its online variant faces significant efficiency bottlenecks due to costly real-time preference sampling and the reward model annotation. |
| Approach: | They propose a framework that transforms static datasets into dynamically adaptive equivalents without the need for an explicit reward model. |
| Outcome: | The proposed approach matches or exceeds the performance of a fully online DPO. |
Diversity Explains Inference Scaling Laws: Through a Case Study of Minimum Bayes Risk Decoding (2025.acl-long)
Copied to clipboard
| Challenge: | Existing studies show that MBR decoding improves model generation performance . however, the theoretical underpinnings of these results remain uncertain . |
| Approach: | They propose a theoretical interpretation of MBR decoding from the perspective of bias–diversity decomposition. |
| Outcome: | The proposed method improves the quality estimation of hypotheses by decomposing bias and diversity into two main factors. |
Synth-SBDH: A Synthetic Dataset of Social and Behavioral Determinants of Health for Clinical Text (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing SBDH datasets lack detailed annotations and are limited in their availability and coverage. |
| Approach: | They propose a synthetic SBDH annotation dataset with detailed SBDH status, temporal information, and rationale across 15 categories. |
| Outcome: | The proposed dataset outperforms models with no Synth-SBDH training on three tasks using real-world clinical datasets from two distinct hospital settings. |
CARO: Chain-of-Analogy Reasoning Optimization for Robust Content Moderation (2026.findings-acl)
Copied to clipboard
| Challenge: | Current large language models struggle with ambiguous content moderation cases due to misleading "decision shortcuts" . authors propose a two-stage training framework to induce robust analogical reasoning in LLMs . |
| Approach: | They propose a two-stage training framework to induce robust analogical reasoning in LLMs . they bootstrap analogy reasoning chains via retrieval-augmented generation and SFT . |
| Outcome: | The proposed framework outperforms state-of-the-art reasoning models and specialized moderation models on ambiguous moderation benchmarks. |
Triple-R: Automatic Reasoning for Fact Verification Using Language Models (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing methods for fact-checking lack external sources and human-understandable explanations for decision-making . existing methods lack external knowledge sources and explanations . |
| Approach: | They propose a framework that uses the Web as an external knowledge source to retrieve relevant evidence for claims and generates reasons based on the retrieved evidence for datasets lacking explanations. |
| Outcome: | The proposed method improves the transparency and interpretability of fact-checking systems by providing human-understandable explanations for decision-making processes. |
How to Protect Yourself from 5G Radiation? Investigating LLM Responses to Implicit Misinformation (2025.emnlp-main)
Copied to clipboard
| Challenge: | Current studies evaluate LLMs on explicit false statements, overlooking how misinformation manifests subtly as unchallenged premises in real-world interactions. |
| Approach: | They propose to use EchoMist to analyze implicit misinformation from diverse sources . they also investigate two mitigation methods to enhance LLMs’ capability to counter implicit mis information. |
| Outcome: | The proposed model fails to detect false premises and generate counterfactual explanations. |
Toward Efficient Sparse Autoencoder-Guided Steering for Improved In-Context Learning in Large Language Models (2025.emnlp-main)
Copied to clipboard
| Challenge: | Sparse autoencoders (SAEs) have emerged as a powerful analytical tool in mechanistic interpretability for large language models (LLMs). |
| Approach: | They propose a novel approach that leverages SAEs to enhance the general in-context learning performance of large language models (LLMs). |
| Outcome: | The proposed method yields a 3.5% improvement across diverse text classification tasks and exhibits greater robustness to hyperparameter variations compared to standard steering approaches. |
TestNUC: Enhancing Test-Time Computing Approaches and Scaling through Neighboring Unlabeled Data Consistency (2025.acl-long)
Copied to clipboard
Henry Peng Zou, Zhengyao Gu, Yue Zhou, Yankai Chen, Weizhi Zhang, Liancheng Fang, Yibo Wang, Yangning Li, Kay Liu, Philip S. Yu
| Challenge: | Test-time computing approaches that leverage additional computational resources during inference have been proven effective in enhancing large language model performance. |
| Approach: | They propose a linearly scaling approach that leverages local consistency of neighboring unlabeled data to improve test-time predictions. |
| Outcome: | The proposed approach outperforms baseline methods such as prompting and self-consistency across eight datasets and performs robustly across embedding models. |
De-Anonymization at Scale via Tournament-Style Attribution (2026.acl-long)
Copied to clipboard
| Challenge: | Large language models (LLMs) are rapidly gaining widespread adoption in real-world use . authors propose a method for attributing authorship among tens of thousands of candidate texts . |
| Approach: | They propose a large-language-model-based method for attributing authorship among tens of thousands of candidate texts. |
| Outcome: | The proposed method improves accuracy and ranking precision over previous approaches. |
Quantifying and Understanding Uncertainty in Large Reasoning Models (2026.acl-long)
Copied to clipboard
| Challenge: | Existing methods for estimating generation uncertainty do not provide finite-sample guarantees for reasoning-answer generation. |
| Approach: | They propose a method that provides the uncertainty of the reasoning-answer structure with statistical guarantees. |
| Outcome: | The proposed method disentangles reasoning quality from answer correctness while establishing theoretical guarantees for efficient explanation methods. |
LCES: Zero-shot Automated Essay Scoring via Pairwise Comparisons Using Large Language Models (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing approaches to automate essay scoring rely on LLMs to generate absolute scores . however, these methods diverge from human evaluations due to model biases and inconsistent scoring . |
| Approach: | They propose a method that formulates AES as a pairwise comparison task using large language models. |
| Outcome: | The proposed method outperforms conventional zero-shot methods in accuracy while maintaining computational efficiency. |
The Arabic Generality Score: Another Dimension of Modeling Arabic Dialectness (2025.emnlp-main)
Copied to clipboard
| Challenge: | Recent work addresses this issue by modeling dialectness as a continuous variable . however, ALDi reduces complex variation to a single dimension . |
| Approach: | They propose a way to model Arabic dialectness as a continuous variable . they propose etymology-aware edit distance and a regression model to model AGS . |
| Outcome: | The proposed approach outperforms baselines on a multi-dialect benchmark. |
ReFreeKV: Towards Threshold-Free KV Cache Compression (2026.findings-acl)
Copied to clipboard
| Challenge: | Towards the KV cache efficiency, we propose a new objective that lifts the threshold constraints for robust KV compression. |
| Approach: | They propose a method that adjusts KV cache budgets while preserving full-cache performance. |
| Outcome: | The proposed method can reduce memory consumption while preserving full-cache performance. |
Safety Sidecar: Reflection-Driven Runtime Control for Safer Agents (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing safety controls fail to provide runtime intervention or cross-architecture portability for autonomous LLM agents. |
| Approach: | They propose a model-agnostic, plug-and-play module to provide arbitrary agent safety control and auditability. |
| Outcome: | The proposed module improves the secure-solution rate by 2.9–11.2 percentage points . it adds only 3.2s to end-to-end latency and a negligible average cost of 5.37 10-4 per scenario . |
R-Fairness: Assessing Fairness of Ranking in Subjective Data (2025.acl-long)
Copied to clipboard
| Challenge: | Subjective data, reflecting individual opinions, permeates platforms like Yelp and Amazon . despite the prevalence of such platforms, little attention has been given to fairness in their context . |
| Approach: | They propose a fairness assessment pipeline that starts with data collection phase and then iterates through rated items. |
| Outcome: | The proposed approach favors groups writing best-ranked reviews over others on collaborative rating platforms. |
SMARTER: A Data-efficient Framework to Improve Toxicity Detection with Explanation via Self-augmenting Large Language Models (2026.acl-long)
Copied to clipboard
| Challenge: | Toxic content encompasses a wide spectrum of terminologies whose definitions vary by platform. |
| Approach: | They propose a 2-stage framework for explainable content moderation using Large Language Models (LLMs) they leverage LLMs’ own outputs to generate synthetic explanations for correct and incorrect labels . they refine explanation quality through cross-model training, allowing weaker models to align with stronger ones. |
| Outcome: | Experiments on 3 benchmarks show that the proposed framework achieves 13% macro-F1 improvement over few-shot baselines using only 6-57% of training data. |
LLM-Powered Benchmark Factory: Reliable, Generic, and Efficient (2026.acl-long)
Copied to clipboard
Peiwen Yuan, Shaoxiong Feng, Yiwei Li, Xinglin Wang, Yueqi Zhang, Jiayi Shi, Chuyi Tan, Boyuan Pan, Yao Hu, Kan Li
| Challenge: | Using generic and efficient benchmark generators, human annotators are limited by inefficiency . current benchmark generator methods rely on seed signals, leading to long cycles and high costs . |
| Approach: | They propose a framework to evaluate LLMs as generic benchmark generators and integrate them as BenchMaker. |
| Outcome: | The proposed framework achieves comparable performance to human-annotated benchmarks on most metrics. |
Controlled Generation for Private Synthetic Text (2025.emnlp-main)
Copied to clipboard
| Challenge: | Text anonymization is essential for developing and deploying AI in high stakes domains . tools for redacting directly identifying content are unlikely to guarantee 100% recall . |
| Approach: | They propose a method for privacy-preserving synthetic text generation that leverages HIPS theory and de-identification principles. |
| Outcome: | The proposed method achieves a strong balance between privacy protection and utility on legal and clinical datasets. |
Prompting Across Time: Evaluating LLMs on Historical and Contemporary Offensive Language (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing research on large language models and hate speech detection has focused on contemporary data. |
| Approach: | They propose to use a modular prompt design to evaluate early-modern English invectives . they propose to widen the scope of NLP research on hate speech beyond the contemporary domain . |
| Outcome: | The proposed model outperforms a modern hate-speech benchmark on Early Modern English invectives . the results show that the model is more robust to contextual and contextual factors than the current model . |
No Need for Explanations: LLMs can implicitly learn from mistakes in-context (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing literature assumes that correct answers to large language models must be accompanied by comprehensive rationales to be helpful. |
| Approach: | They propose to show incorrect answers to Large Language Models (LLMs) as a popular strategy to improve their performance in reasoning-intensive tasks. |
| Outcome: | The proposed approach outperforms chain-of-thought prompting in math reasoning tasks. |
InsAT: Instance-aware Semantic Alignment and Transfer from Human–Object Keypoints for Zero-to-Few-shot Action Understanding (2026.acl-long)
Copied to clipboard
| Challenge: | Existing zero-shot (ZS) approaches emphasize human motion while underutilizing contextual information, particularly human–object interactions. |
| Approach: | They propose a framework for ZS recognition and zero-to-few-shot adaptation that leverages instance-level language descriptions. |
| Outcome: | The proposed framework outperforms keypoint-based ZS methods while remaining data-efficient and robust. |
CSRP: Chain-of-Thought Reasoning for Chinese Text Correction via Reinforcement Learning with Efficiency-Aware Rewards (2026.acl-long)
Copied to clipboard
| Challenge: | Large Language Models lack specialized priors for subtle grammatical distinctions, and Supervised Fine-Tuning fails to optimize for precision-focused metrics. |
| Approach: | They propose a framework that builds correction capability through Continual Pre-training on 5.9M balanced samples to internalize domain knowledge. |
| Outcome: | The proposed framework outperforms existing models on the NACGEC benchmark with 50.99 F0.5 and 57.17 precision while mitigating over-correction bias. |
From Where Words Come: Efficient Regularization of Code Tokenizers Through Source Attribution (2026.acl-long)
Copied to clipboard
| Challenge: | Currently, subword tokenization is the most common approach for vocabulary building in large models. |
| Approach: | They propose to regularize training and minimize overfitting by using source-attributed BPE . they find that undertrained tokens are prone to producing unused, unusable tokens . |
| Outcome: | The proposed techniques reduce the number of under-trained tokens while maintaining the same inference procedure as with regular BPE. |
Too Long, Do Re-weighting for Efficient LLM Reasoning Compression (2026.acl-long)
Copied to clipboard
Zhong-Zhi Li, Xiao Liang, Zihao Tang, Lei Ji, Peijie Wang, Haotian Xu, Xing W, Haizhen Huang, Weiwei Deng, Yeyun Gong, Zhijiang Guo, Xiao Liu, Fei Yin, Cheng-Lin Liu
| Challenge: | Large Language Models (LLMs) have recently achieved remarkable progress on complex reasoning tasks by leveraging extended Chain-of-Thought (CoT) techniques. |
| Approach: | They propose a method that uses Extended Chain-of-Thought (EFT) to reduce the number of output tokens by nearly 40% while maintaining the accuracy of the reasoning. |
| Outcome: | The proposed method reduces the number of output tokens by nearly 40% while maintaining the accuracy of the reasoning. |
Interpretability from the Ground Up: Stakeholder-Centric Design of Automated Scoring in Educational Assessments (2026.findings-acl)
Copied to clipboard
| Challenge: | Despite increasing demand for transparency and interpretability, the field has yet to develop a widely accepted solution for interpretable automated scoring to be used in large-scale real-world assessments. |
| Approach: | They propose to develop four principles of interpretability targeted at assessment stakeholder groups to address the need for transparency and interpretability in automated scoring. |
| Outcome: | The proposed framework outperforms many uninterpretable scoring methods in terms of scoring accuracy and is, on average, within 0.06 QWK of the uninterprétable SOTA. |
Single-Agent Generation Surpasses Multi-Agent Systems in Semantic Diversity (2026.findings-acl)
Copied to clipboard
| Challenge: | Multi-Agent Systems (MAS) are used to improve reasoning diversity and robustness by simulating interactions among agents with distinct roles. |
| Approach: | They find that a Multi-Output strategy produces the highest diversity without degrading logical validity. |
| Outcome: | The proposed approach outperforms multi-agent systems in semantic diversity . the results point to a more efficient and effective way to expand diversity - the authors say . |
LLM4Cell: Taxonomy and Evaluation of LLM and Agentic Models for Single-Cell Biology (2026.acl-long)
Copied to clipboard
Sajib Acharjee Dip, Adrika Zafor, Bikash Kumar Paul, Uddip Acharjee Shuvo, Muhit Islam Emon, Xuan Wang, Liqing Zhang
| Challenge: | Large language models are transforming biomedical discovery by linking molecular patterns with knowledge encoded in text. |
| Approach: | They propose to map 58 foundation and agentic models developed for single-cell research into eight key analytical tasks. |
| Outcome: | The proposed models are applied to eight key analytical tasks including annotation, trajectory inference, perturbation modeling, and drug-response prediction. |
Reasoning Traces Shape Outputs but Models Won’t Say So (2026.acl-long)
Copied to clipboard
| Challenge: | Large reasoning models (LRMs) generate explicit reasoning traces before producing answers, offering a window into their decisionmaking. |
| Approach: | They propose a method that injects synthetic reasoning snippets into a model’s reasoning trace and measures whether the model follows the injected reasoning and acknowledges doing so. |
| Outcome: | The proposed method reveals that models refuse to disclose their influence when asked to explain their changed answers. |
Credal Concept Bottleneck Models for Epistemic–Aleatoric Uncertainty Decomposition (2026.acl-long)
Copied to clipboard
| Challenge: | Existing models face challenges when dealing with uncertainty. |
| Approach: | They propose a framework that decomposes concept uncertainty by construction . epistemic uncertainty is positively associated with prediction errors, whereas aleatoric uncertainty closely tracks disagreement . |
| Outcome: | The proposed framework decomposes concept uncertainty by construction . epistemic uncertainty is positively associated with prediction errors, whereas aleatoric uncertainty closely tracks disagreement . |
Learning Faster with Better Tokens: Parameter-Efficient Vocabulary Adaptation for Specialized Text Summarization (2026.acl-long)
Copied to clipboard
| Challenge: | a new approach to adapt generalist models to expert domains is needed to overcome this problem. |
| Approach: | They propose a parameter-efficient domain adaptation approach that combines vocabulary adaptation with pretraining for LLM-based text summarization. |
| Outcome: | The proposed approach reduces training time by 35-55% over continual pretraining and reduces parameter counts up to 37% w.r.t expansion-only methods. |
Reinforced Efficient Reasoning via Semantically Diverse Exploration (2026.acl-long)
Copied to clipboard
Ziqi Zhao, Zhaochun Ren, Jiahong Zou, Liu Yang, Zhiwei Xu, Xuri Ge, Zhumin Chen, Xinyu Ma, Daiting Shi, Shuaiqiang Wang, Dawei Yin, Xin Xin
| Challenge: | Existing methods for reinforcement learning with verifiable rewards suffer from limited exploration diversity and inefficient reasoning. |
| Approach: | They propose a method that rewards concise and correct reasoning while penalizing unnecessarily long reasoning chains. |
| Outcome: | Extensive experiments on Qwen and Llama models validate the effectiveness and efficiency of ROSE. |