Papers by Di Lu
Less is More: Pretrain a Strong Siamese Encoder for Dense Text Retrieval Using a Weak Decoder (2021.emnlp-main)
Copied to clipboard
Shuqi Lu, Di He, Chenyan Xiong, Guolin Ke, Waleed Malik, Zhicheng Dou, Paul Bennett, Tie-Yan Liu, Arnold Overwijk
| Challenge: | Dense retrieval requires high-quality text sequence embeddings to support effective search in the representation space. |
| Approach: | They propose a self-learning method that pre-trains the autoencoder using a weak decoder to push the encoder to provide better sequence representations. |
| Outcome: | The proposed model significantly boosts the effectiveness and few-shot ability of dense retrieval models on web search, news recommendation, and open domain question answering. |
Cross-media Structured Common Space for Multimedia Event Extraction (2020.acl-main)
Copied to clipboard
| Challenge: | We propose a new task to extract events and their arguments from multimedia documents . traditional methods target text, images or videos, but multimedia content is distributed via multimedia . |
| Approach: | They propose a method that encodes structured representations of semantic information from textual and visual data into a common embedding space. |
| Outcome: | The proposed method achieves 4.0% and 9.8% absolute gains on text event argument role labeling and visual event extraction. |
Cross-lingual Structure Transfer for Relation and Event Extraction (D19-1)
Copied to clipboard
| Challenge: | Existing approaches to identify complex semantic structures are difficult to train from under-annotated sources. |
| Approach: | They exploit relation- and event-relevant language-universal features to train relation or event extractors from source annotations and apply them to target languages. |
| Outcome: | The proposed approach achieves comparable performance to state-of-the-art models trained on 3,000 manually annotated mentions. |
The ApposCorpus: a new multilingual, multi-domain dataset for factual appositive generation (2020.coling-main)
Copied to clipboard
| Challenge: | appositives are phrases that appear next to a noun phrase and serve an explicative function. |
| Approach: | They propose a more realistic end-to-end definition of appositive generation with a dataset that spans four languages and two entity types. |
| Outcome: | The proposed model is non-trivial and leaves plenty of room for improvement. |
Curriculum-RLAIF: Curriculum Alignment with Reinforcement Learning from AI Feedback (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to align large language models with human preferences are limited in generalizability due to distribution shift, preference label noise, and mismatch of challenging samples with model capacity. |
| Approach: | They propose a framework that constructs preference pairs with varying difficulty levels and then produces a specific curriculum for reward model training. |
| Outcome: | The proposed framework improves generalizability of reward models by a significant margin without incurring additional inference costs compared to existing non-curriculum baselines. |
Platforms for Non-speakers Annotating Names in Any Language (P18-4)
Copied to clipboard
| Challenge: | Traditionally, native speakers of a language have been asked to annotate a corpus in that language. |
| Approach: | They propose two annotation platforms that allow an English speaker to annotate names for any language without knowing the language. |
| Outcome: | The proposed annotations achieved state-of-the-art performance on two surprise languages and ten languages at TAC-KBP EDL2017. |
TaCIE: Enhancing Instruction Comprehension in Large Language Models through Task-Centred Instruction Evolution (2025.coling-main)
Copied to clipboard
| Challenge: | Existing methods for fine-tuning Large Language Models (LLMs) encounter performance limitations, impeding further enhancements in code generation tasks. |
| Approach: | They propose to combine two distinct prompts through a hybridization process to enhance the evolution of training prompts for code LLMs. |
| Outcome: | The proposed method significantly improves the performance of Code LLMs across five code generation benchmarks. |
Entity-aware Image Caption Generation (D18-1)
Copied to clipboard
| Challenge: | Existing image captioning approaches generate generic descriptions of visual content and ignore background information. |
| Approach: | They propose a task which generates informative image captions using images and hashtags as input. |
| Outcome: | The proposed model outperforms unimodal baselines significantly with evaluation metrics on a dataset from Flickr. |
AgentFactory: A Self-Evolving Framework Through Executable Subagent Accumulation and Reuse (2026.acl-demo)
Copied to clipboard
| Challenge: | Existing frameworks for building LLM-based agents treat agent behavior as static-knowledge gained during execution is not preserved for future use. |
| Approach: | They propose a new paradigm that preserves successful task solutions as executable subagent code rather than textual experience. |
| Outcome: | The proposed agent-based agent-driven paradigm preserves successful tasks as executable subagent code rather than textual experience. |
ShieldLM: Empowering LLMs as Aligned, Customizable and Explainable Safety Detectors (2024.findings-emnlp)
Copied to clipboard
Zhexin Zhang, Yida Lu, Jingyuan Ma, Di Zhang, Rui Li, Pei Ke, Hao Sun, Lei Sha, Zhifang Sui, Hongning Wang, Minlie Huang
| Challenge: | Existing tools for detecting safety issues in LLMs are expensive and inefficient. |
| Approach: | They propose an LLM-based safety detector which annotates the safety of queries and provides explanations for its decisions. |
| Outcome: | The proposed detector outperforms baselines on four sets of query-response pairs and is effective as a safety evaluator for advanced LLMs. |
Event Extraction as Question Generation and Answering (2023.acl-short)
Copied to clipboard
| Challenge: | Recent work on Event Extraction addresses the error propagation issue found in token-based classification approaches. |
| Approach: | They propose a Question Generation (QG) model that generates questions that leverage contextual information instead of fixed templates. |
| Outcome: | The proposed model outperforms all previous single-task-based models on the ACE05 English dataset. |
Parameter Importance is Not Static: Evolving Parameter Isolation for Supervised Fine-Tuning (2026.acl-long)
Copied to clipboard
Zekai Lin, Chao Xue, Di Liang, Xingsheng Han, Peiyang Liu, Xianjie Wu, Lei Jiang, Yu Lu, Bob Simons, Shuang Liang, Minlong Peng
| Challenge: | Recent approaches to fine-tuning of large language models suffer from task interference and catastrophic forgetting. |
| Approach: | They propose a fine-tuning framework that adapts isolation decisions based on online estimates of parameter importance. |
| Outcome: | The proposed framework reduces interference and forgetting while releasing outdated parameters to recover plasticity. |
D-Artemis: A Deliberative Cognitive Framework for Mobile GUI Multi-Agents (2026.findings-acl)
Copied to clipboard
Hongze Mi, Yibo Feng, WenJie Lu, Yuqi Wang, Jinyuan Li, Song Cao, He Cui, Tengfei Tian, Xuelin Zhang, Haotian Luo, Di Sun, Jun Fang, Hua Chai, Naiqiang Tan, Gang Pan
| Challenge: | Graphical User Interface (GUI) agents aim to automate a wide spectrum of human tasks by emulating user interaction. |
| Approach: | They propose a deliberative framework that leverages a fine-grained tip retrieval mechanism to inform its decision-making process. |
| Outcome: | The proposed framework achieves SOTA among open-source general models on AndroidWorld and ScreenSpot-V2 . it leverages a fine-grained, app-specific tip retrieval mechanism to inform its decision-making process . |
Why Supervised Fine-Tuning Fails to Learn: A Systematic Study of Incomplete Learning in Large Language Models (2026.acl-long)
Copied to clipboard
Chao Xue, Yao Wang, Mengqiao Liu, Di Liang, Xingsheng Han, Peiyang Liu, Xianjie Wu, Chenyao Lu, Lei Jiang, Yu Lu, Haibo Shi, Shuang Liang, Minlong Peng, Flora D. Salim
| Challenge: | Incomplete learning is widespread and heterogeneous in large language models . authors identify five recurrent sources of incomplete learning: missing prerequisite knowledge, conflicts between SFT supervision and pre-training knowledge, internal inconsistencies within SFT data, left-side forgetting during sequential fine-tuning, and insufficient optimization for rare or complex patterns. |
| Approach: | They propose a diagnostic-first framework that maps incomplete learning to causes . they identify five recurrent sources of incomplete learning: missing prerequisite knowledge, conflicts between supervision and pre-training knowledge, internal inconsistencies, left-side forgetting during sequential fine-tuning, and insufficient optimization for rare or complex patterns. |
| Outcome: | The proposed framework maps incomplete learning to causes using observable training and inference signals. |
ELISA-EDL: A Cross-lingual Entity Extraction, Linking and Localization System (N18-5)
Copied to clipboard
| Challenge: | ELISA-EDL is a cross-lingual entity extraction, linking and localization system for Wikipedia languages. |
| Approach: | They propose a cross-lingual entity extraction, linking and localization system for English speakers . it extracts entities from unstructured text in any of 282 Wikipedia languages and links them to English knowledge bases . |
| Outcome: | The proposed system extracts entity mentions from Wikipedia and links them to English knowledge bases and visualizes locations related to disaster topics on a world heatmap. |
BUMP: A Benchmark of Unfaithful Minimal Pairs for Meta-Evaluation of Faithfulness Metrics (2023.acl-long)
Copied to clipboard
Liang Ma, Shuyang Cao, Robert L Logan IV, Di Lu, Shihao Ran, Ke Zhang, Joel Tetreault, Alejandro Jaimes
| Challenge: | Existing benchmarks measure the correlation with human judgements of faithfulness on model-generated summaries, but they are insufficient for diagnosing whether metrics are consistent, effective on human-written texts, and sensitive to different error types. |
| Approach: | They propose to use unfaithful minimal pairs to measure the consistency of automatic faithfulness metrics by comparing human-written summary pairs with a dataset of 889 human-writing, minimally different summary pairs. |
| Outcome: | The proposed benchmarks show that the most discriminative metrics tend not to be the most consistent, and that the best performing metrics are sensitive to errors. |
RoleMRC: A Fine-Grained Composite Benchmark for Role-Playing and Instruction-Following (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing role-playing datasets mostly contribute to controlling role style and knowledge boundaries, but overlook role-following in instruction-follower scenarios. |
| Approach: | They propose a fine-grained role-playing and instruction-following composite benchmark, named RoleMRC, which includes multi-turn dialogues between ideal roles and humans, including free chats or discussions upon given passages . |
| Outcome: | The proposed model improves instruction-following without compromising general role-playing and reasoning capabilities. |
WaterSeeker: Pioneering Efficient Detection of Watermarked Segments in Large Documents (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods focus on distinguishing fully watermarked text from non-watermarked text, overlooking real-world scenarios where LLMs generate only brief segments within longer documents. |
| Approach: | They propose a method to detect watermarked segments in large documents using an anomaly extraction method and a local traversal. |
| Outcome: | The proposed method achieves a superior balance between detection accuracy and computational efficiency. |
Cross-lingual Structure Transfer for Zero-resource Event Extraction (2020.lrec-1)
Copied to clipboard
| Challenge: | Existing approaches for information extraction only use name tagging . Currently, most successful cross-lingual transfer learning methods are limited to sequence labeling . |
| Approach: | They propose a share-and-transfer framework to transfer graph structures across languages . they propose to convert sentences in any language to language-universal graph structures . |
| Outcome: | The proposed framework performs comparable to state-of-the-art models on three languages without annotations. |
Tell Me What You Don’t Know: Enhancing Refusal Capabilities of Role-Playing Agents via Representation Space Analysis and Editing (2025.findings-acl)
Copied to clipboard
Wenhao Liu, Siyu An, Junru Lu, Muling Wu, Tianlong Li, Xiaohua Wang, Changze Lv, Xiaoqing Zheng, Di Yin, Xing Sun, Xuanjing Huang
| Challenge: | Role-playing Agents (RPAs) struggle to recognize and respond to hard queries that conflict with their role-play knowledge. |
| Approach: | They propose a lightweight representation editing approach that conveniently shifts conflicting requests to the rejection region, thereby enhancing the model’s refusal accuracy. |
| Outcome: | The proposed model improves RPAs’ refusal ability of conflicting requests while maintaining their general role-playing capabilities. |
Reason Only When Needed: Efficient Generative Reward Modeling via Model-Internal Uncertainty (2026.findings-acl)
Copied to clipboard
Chao Xue, Yao Wang, Mengqiao Liu, Di Liang, Xingsheng Han, Peiyang Liu, Xianjie Wu, Chenyao Lu, Lei Jiang, Yu Lu, Haibo Shi, Shuang Liang, Minlong Peng, Flora D. Salim
| Challenge: | Existing approaches to generating reward models rely on voting-based mechanisms to evaluate CoT outputs. |
| Approach: | They propose an efficient generative reward modeling framework grounded in model-internal uncertainty. |
| Outcome: | The proposed framework reduces inference cost while improving answer accuracy. |
DialogBench: Evaluating LLMs as Human-like Dialogue Systems (2024.naacl-long)
Copied to clipboard
| Challenge: | Existing benchmarks only evaluate LLMs' abilities for task completion as assistant AI. |
| Approach: | They propose a dialogue evaluation benchmark that contains 12 dialogue tasks to evaluate LLMs' capabilities as human-like dialogue systems. |
| Outcome: | The proposed benchmark contains 12 tasks to evaluate LLMs' capabilities . it shows that instruction tuning improves human likeness, but not as human-like systems . |
Eliminating Biased Length Reliance of Direct Preference Optimization via Down-Sampled KL Divergence (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing studies attributed verbosity to biased labels, but new research shows that DPO can be effective in mitigating verboses. |
| Approach: | They propose to use a method to reduce the amount of verbosity in LLMs by using a downsampling approach. |
| Outcome: | The proposed approach overcomes the problem of verbosity by reducing the length reliance of the proposed algorithm. |
PerfCoder: Large Language Models for Interpretable Code Performance Optimization (2026.findings-acl)
Copied to clipboard
Jiuding Yang, Shengyao Lu, Hongxuan Liu, Shayan Shirahmad Gale Bagi, Zahra Fazel, Tomasz Czajkowski, Di Niu
| Challenge: | Large language models (LLMs) have advanced automatic code generation, but their ability to produce high-performance code remains limited. |
| Approach: | They propose a family of large language models that generate performance-enhanced code through interpretable and customized optimization strategies. |
| Outcome: | The proposed model outperforms existing models on the PIE code performance benchmark and produces interpretable feedback that can guide larger LLMs in a planner–optimizer workflow. |
Separate the Wheat from the Chaff: A Post-Hoc Approach to Safety Re-Alignment for Fine-Tuned Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models achieve effective safety alignment at the time of release, but fine-tuning often compromises safety mechanisms. |
| Approach: | They propose a method that performs safety realignment for large language models . they identify unsafe delta parameters from the fine-tuned models and recalibrate the retained parameters . |
| Outcome: | The proposed method improves safety performance on safety benchmarks and jailbreak attacks while maintaining their performance on downstream tasks. |
A New Task and Dataset on Detecting Attacks on Human Rights Defenders (2023.findings-acl)
Copied to clipboard
| Challenge: | Existing datasets for event extraction cannot extract information from textual sources. |
| Approach: | They propose to use crowdsourced annotations on 500 online news articles to train and evaluate baseline models to predict annotated characteristics. |
| Outcome: | The proposed dataset includes crowdsourced annotations on 500 online news articles and includes fine-grained information about the type and location of attacks, as well as information about victims. |
FIPO: Free-form Instruction-oriented Prompt Optimization with Preference Dataset and Modular Fine-tuning Schema (2025.coling-main)
Copied to clipboard
| Challenge: | naive prompts can enhance the task performance of large language models, but they are resource-intensive. |
| Approach: | They propose an automatic prompt optimization method that refines naive prompts according to task outputs from in-box testing models. |
| Outcome: | The proposed method is based on a large-scale dataset and performed fairly across multiple models. |
SlotRefine: A Fast Non-Autoregressive Model for Joint Intent Detection and Slot Filling (2020.emnlp-main)
Copied to clipboard
| Challenge: | Slot filling and intent detection are two main tasks in spoken language understanding systems. |
| Approach: | They propose a non-autoregressive slot filling model with two-pass iteration mechanism to handle uncoordinated slots problem. |
| Outcome: | The proposed model significantly outperforms previous models in slot filling task while speeding up decoding. |
Olá, Bonjour, Salve! XFORMAL: A Benchmark for Multilingual Formality Style Transfer (2021.naacl-main)
Copied to clipboard
| Challenge: | XFORMAL benchmarks formal reformulations of informal text in Brazilian Portuguese, French, and Italian . most work on style transfer within English, while covering different languages has received disproportional interest. |
| Approach: | They create a benchmark of multiple formal reformulations of informal text in Brazil, Brazil, and Italy. |
| Outcome: | XFORMAL benchmarks formal reformulations of informal text in Brazilian Portuguese, French, and Italian . results show that state-of-the-art approaches perform close to simple baselines . |
Error Analysis of Uyghur Name Tagging: Language-specific Techniques and Remaining Challenges (L18-1)
Copied to clipboard
| Challenge: | despite efforts at name tagging, there is limited understanding on the performance ceiling . despite the high-resource language, there are very few natural language processing tools available . |
| Approach: | They propose to use a machine learning model to identify Uyghur name tagger errors . they conclude that such a model is unlikely to be effective for Uygur, or low-resource languages . |
| Outcome: | The proposed model is unlikely to be effective for Uyghur, or low-resource languages in general, the authors argue . they show that the proposed model can be used for high-res languages with superficial features . |
Double Path Networks for Sequence to Sequence Learning (C18-1)
Copied to clipboard
| Challenge: | Existing approaches for Sequence to Sequence learning have been developed . convolutional neural networks and self-attention networks are the most popular . |
| Approach: | They propose to integrate convolutional and self-attention layers into a double path network for sequence to sequence learning. |
| Outcome: | The proposed method significantly improves performance over state-of-the-art systems. |
Visual Attention Model for Name Tagging in Multimodal Social Media (P18-1)
Copied to clipboard
| Challenge: | Name tagging is a key task for language understanding, but is often limited by the short textual components. |
| Approach: | They propose a novel model architecture based on visual attention that outperforms other methods . they use multimodal datasets to analyze the name tagging task on social media . |
| Outcome: | The proposed model outperforms existing methods and significantly outperformed existing methods. |
Multi-View Source Ablation for Faithful Summarization (2023.findings-eacl)
Copied to clipboard
| Challenge: | MuFaSSa is a metric for evaluating faithfulness of abstractive summaries . it uses different strategies to remove information from source document to form multiple ablated views . |
| Approach: | They propose a metric for evaluating faithfulness of abstractive summaries using multiple ablated views. |
| Outcome: | The proposed metric outperforms existing models on summarization tasks and human-annotated faithfulness labels. |
New Terms, New Toxicity: Consensus-based Chinese Neologism Toxicity Detection via Search-Augmented LLMs (2026.acl-long)
Copied to clipboard
Shiyao Cui, QingLin Zhang, Di Wang, Yida Lu, Zhexin Zhang, Jinhua Gao, Jinglin Yang, Min He, Han Qiu, Minlie Huang
| Challenge: | Neologisms can foster new linguistic consensus by stabilizing shared meanings and usage in common communicative norms. |
| Approach: | They propose a taxonomy that captures the origins and consensus-verification criteria of toxic neologisms . they propose 'SeTox' framework that integrates real-time web context for naeologim detection . |
| Outcome: | The proposed framework outperforms large-scale models in detecting neologism toxicity. |
Decoder-Only LLMs can be Masked Auto-Encoders (2025.acl-short)
Copied to clipboard
Dan Qiao, Yuan Gao, Zheming Yang, Di Yang, Ziheng Wu, Pengcheng Lu, Minghui Qiu, Juntao Li, Min Zhang
| Challenge: | Modern NLP workflows require different models for generation and embedding tasks. |
| Approach: | They propose a method that transforms an LLM into a Uni-Directional Masked Auto-Encoder. |
| Outcome: | The proposed method achieves state-of-the-art under unsupervised conditions with merely 100 training steps. |
Improving Visual-Semantic Embedding with Adaptive Pooling and Optimization Objective (2023.eacl-main)
Copied to clipboard
Zijian Zhang, Chang Shu, Ya Xiao, Yuan Shen, Di Zhu, Youxin Chen, Jing Xiao, Jey Han Lau, Qian Zhang, Zheng Lu
| Challenge: | Recent VSE models combine simple pooling methods with hard triplet loss to improve performance. |
| Approach: | They propose an adaptive pooling strategy that allows the model to learn how to aggregate features through a combination of simple pooling methods. |
| Outcome: | The proposed strategy outperforms current state-of-the-art systems on image-to-text and text-toimage retrieval. |
ASR-EC Benchmark: Evaluating Large Language Models on Chinese ASR Error Correction (2025.emnlp-industry)
Copied to clipboard
| Challenge: | Automatic Speech Recognition (ASR) systems have a substantial number of erroneous recognition due to environmental noise, ambiguity, etc. |
| Approach: | They use a benchmark dataset to analyze ASR errors in the Chinese language . they then apply large language models to correct ASR error correction . |
| Outcome: | The proposed method is based on a dataset of ASR errors in the Chinese language . it shows prompting is not effective for ASR error correction . |
CEHA: A Dataset of Conflict Events in the Horn of Africa (2025.coling-main)
Copied to clipboard
Rui Bai, Di Lu, Shihao Ran, Elizabeth M. Olson, Hemank Lamba, Aoife Cahill, Joel Tetreault, Alejandro Jaimes
| Challenge: | Existing datasets categorizing conflict events do not cover all of the fine-grained types of conflict relevant to areas like the Horn of Africa. |
| Approach: | They propose to use online news articles to categorize violent conflict events . they propose to extract event-relevance and event-types from 500 English event descriptions . |
| Outcome: | The proposed dataset categorizes conflict risk according to specific areas required by stakeholders in the Humanitarian-Peace-Development Nexus. |