Papers with textual representations
Lexicon-Level Contrastive Visual-Grounding Improves Language Modeling (2024.findings-acl)
Copied to clipboard
| Challenge: | Neural language models (LMs) are trained on orders of magnitude more language data than human language learners receive, but without supervision from other sensory modalities that play a crucial role in human learning. |
| Approach: | They propose a grounded language learning procedure that leverages visual supervision to improve textual representations. |
| Outcome: | The proposed procedure outperforms standard language-only models in terms of learning efficiency in small and developmentally plausible data regimes and improves perplexity by around 5% on multiple language modeling tasks compared to other models trained on the same amount of text data. |
Knowledge Graph Entity Typing with Curriculum Contrastive Learning (2025.coling-main)
Copied to clipboard
| Challenge: | Existing knowledge graphs suffer from incomplete type annotations because they are manually constructed by domain experts. |
| Approach: | They propose a CCLET model using the Curriculum Contrastive Learning strategy for KGET to fuse the entity related semantic and the structural information of the Knowledge Graph (KG) they define the difficulty of the course by controlling the level of added noise and aim to accurately learn with curriculum contrastive learning strategy from easy to difficult. |
| Outcome: | The proposed model outperforms state-of-the-art models and is highly accurate across multiple learning environments. |
AnyGraph: Graph Foundation Model in the Wild (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing graph learning models struggle to extract generalizable insights from heterogeneous graph data, requiring extensive fine-tuning and limiting versatility across domains. |
| Approach: | They propose a graph foundation model that can handle key challenges such as Structure Heterogenity and Feature Heterogenicity. |
| Outcome: | The proposed model can handle key challenges such as structure heterogeneity, Feature heterogenity and fast adaptation across domains. |
MIReAD: Simple Method for Learning High-quality Representations from Scientific Documents (2023.acl-short)
Copied to clipboard
| Challenge: | Pretrained language models can learn rich textual representations, but they cannot provide powerful document-level representations for scientific articles. |
| Approach: | They propose a transformer-based method that learns semantically meaningful representations from scientific papers by fine-tuning transformer models to predict the target journal class based on the abstract. |
| Outcome: | The proposed method outperforms six existing models for representation learning on scientific documents across four evaluation standards. |
Math Word Problem Solving by Generating Linguistic Variants of Problem Statements (2023.acl-srw)
Copied to clipboard
Syed Rifat Raiyan, Md Nafis Faiyaz, Shah Md. Jawad Kabir, Mohsinul Kabir, Hasan Mahmud, Md Kamrul Hasan
| Challenge: | Existing models for solving Math Word Problems depend on shallow heuristics and spurious correlations to derive the solution expressions. |
| Approach: | They propose a framework for MWP solvers based on generation of linguistic variants of problem text. |
| Outcome: | The proposed framework improves the mathematical reasoning and robustness of the proposed model. |
Modeling Language Usage and Listener Engagement in Podcasts (2021.acl-long)
Copied to clipboard
| Challenge: | linguistic factors such as vocabulary diversity, distinctiveness, emotion, and syntax are highly predictive of engagement in podcasts, but little research has been done into how they contribute to overall listener engagement. |
| Approach: | They build models with different textual representations to test popular wisdom about stylistic elements in high-engagement podcasts, corroborating some pieces of advice and adding new perspectives on others. |
| Outcome: | The proposed models show that stylistic features are highly predictive of engagement for podcasts with low absolute streams than for the most popular podcasts. |
LangVAE and LangSpace: Building and Probing for Language Model VAEs (2025.emnlp-demos)
Copied to clipboard
| Challenge: | a novel framework for modular construction of variational autoencoders (VAEs) on pre-trained large language models (LLMs) is presented. |
| Approach: | They propose a modular framework for variational autoencoders on top of pre-trained large language models. |
| Outcome: | The proposed framework can encode pre-trained language models into more compact and semantically disentangled representations. |
Generalizing Word Embeddings using Bag of Subwords (D18-1)
Copied to clipboard
| Challenge: | Existing word embeddings techniques have a fixed vocabulary, i.e., they can only provide vectors over a finite set of common words that appear frequently in a given corpus. |
| Approach: | They propose a subword-level word vector generation model that views words as bags of character n-grams and provides good vectors for rare or unseen words. |
| Outcome: | The proposed model performs state-of-the-art in English word similarity task and in joint prediction of part-of speech tag and morphosyntactic attributes in 23 languages. |
Incorporating Visual Semantics into Sentence Representations within a Grounded Space (D19-1)
Copied to clipboard
| Challenge: | Language grounding is an active field aiming at enriching textual representations with visual information. |
| Approach: | They propose to transfer visual information to textual representations by learning an intermediate representation space: the grounded space. |
| Outcome: | The proposed model outperforms the previous state-of-the-art on classification and semantic relatedness tasks. |
Accurate Text-Enhanced Knowledge Graph Representation Learning (N18-1)
Copied to clipboard
| Challenge: | Existing representation learning methods for knowledge graph representation do not consider the ambiguity of relations and entities. |
| Approach: | They propose a text-enhanced knowledge graph representation learning method which exploits the entity descriptions and triple-specific relation mention to enhance representations. |
| Outcome: | The proposed method outperforms existing representation learning models on link prediction and triple classification tasks and significantly outperformed existing models. |
BioT5+: Towards Generalized Biological Understanding with IUPAC Integration and Multi-task Tuning (2024.findings-acl)
Copied to clipboard
Qizhi Pei, Lijun Wu, Kaiyuan Gao, Xiaozhuan Liang, Yin Fang, Jinhua Zhu, Shufang Xie, Tao Qin, Rui Yan
| Challenge: | BioT5+ is an extension of the BioT5, but lacked a nuanced understanding of molecular structures. |
| Approach: | They propose a new bio-entity modeling framework, BioT5+, which integrates IUPAC names and molecule data. |
| Outcome: | The proposed model bridges the gap between molecular representations and textual descriptions and improves the grounded reasoning of bio-text and bio-sequences. |
Entity Disambiguation with Entity Definitions (2023.eacl-main)
Copied to clipboard
| Challenge: | Entity Disambiguation (ED) is a crucial problem in Natural Language Processing (NLP). |
| Approach: | They propose to use Wikipedia titles as the textual representation of each candidate to improve the generalization capability over unseen patterns. |
| Outcome: | The proposed model improves on 2 out of 6 benchmarks and is generalized over unseen patterns. |
Multimodal Frame Identification with Multilingual Evaluation (N18-1)
Copied to clipboard
| Challenge: | FrameNet Semantic Role Labeling aims to disambiguate situations around predicates using textual representations. |
| Approach: | They extend a frame identification task to leverage multimodal representations to improve FrameNet Semantic Role Labeling. |
| Outcome: | The proposed system outperforms its unimodal counterpart on the English frameNet and its German counterpart on IMAGINED words. |
Prepending or Cross-Attention for Speech-to-Text? An Empirical Comparison (2025.naacl-long)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have been successful in NLP tasks, but there is growing interest in extending their capabilities to speech. |
| Approach: | They propose to use dense feature prepending (DFP) to integrate speech into LLMs to enable end-to-end training with a speech encoder. |
| Outcome: | The proposed approach does not show a clear advantage over cross-attention. |
CODERL+: Improving Code Generation via Reinforcement with Execution Semantics Alignment (2026.acl-long)
Copied to clipboard
Xue Jiang, Yihong Dong, Mengyang Liu, Deng Hongyi, Tian Wang, Yongding Tao, Zhi Jin, Wenpin Jiao, Ge Li
| Challenge: | Large Language Models excel at code generation by learning from vast code corpora, but a fundamental semantic gap remains between training on textual patterns and the goal of functional correctness . reinforcement learning with verifiable rewards (RLVR) approaches are inefficient for establishing a well-aligned connection between the textual representation of code and its execution semantics. |
| Approach: | They propose a novel approach that integrates execution semantics alignment into the RLVR training pipeline for code generation. |
| Outcome: | The proposed model outperforms baseline training and RLVR and shows strong applicability across RL and LLMs. |
Distill Visual Chart Reasoning Ability from LLMs to MLLMs (2025.findings-emnlp)
Copied to clipboard
Wei He, Zhiheng Xi, Wanxu Zhao, Xiaoran Fan, Yiwen Ding, Zifei Shan, Tao Gui, Qi Zhang, Xuanjing Huang
| Challenge: | a new method for generating chart annotations is proposed to improve visual reasoning in multimodal large language models. |
| Approach: | They propose a code-as-intermediary translation method for distilling visual reasoning abilities from LLMs to MLLMs. |
| Outcome: | The proposed method is cost-effective, efficient and scalable. |
Beyond End-to-End VLMs: Leveraging Intermediate Text Representations for Superior Flowchart Understanding (2025.naacl-long)
Copied to clipboard
| Challenge: | Flowcharts are typically presented as images, driving the trend of using vision-language models for end-to-end flowchart understanding. |
| Approach: | They propose a vision-language model (VLM) that generates textual representations from flowchart images and a textual Reasoner that performs question-answering based on the text representations. |
| Outcome: | Experiments on the FlowVQA and FlowLearn benchmarks demonstrate TextFlow’s state-of-the-art performance as well as its robustness. |
Learning Disentangled Textual Representations via Statistical Measures of Similarity (2022.acl-long)
Copied to clipboard
| Challenge: | Existing approaches to disentangle a sensitive attribute from textual representations require training and multiple parameter updates. |
| Approach: | They propose a family of regularizers for learning disentangled representations that do not require training. |
| Outcome: | The proposed regularizers are faster and faster and achieve better results when combined with pretrained and randomly initialized text encoders. |
Filling the Image Information Gap for VQA: Prompting Large Language Models to Proactively Ask Questions (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) demonstrate impressive reasoning ability and the maintenance of world knowledge in natural language tasks. |
| Approach: | They propose a framework that enables LLMs to ask relevant questions to uncover more details in the image, along with filters for refining the generated information. |
| Outcome: | The proposed framework boosts the performance of baseline methods by 2.15% on OK-VQA and achieves consistent improvements across different LLMs. |
t-HNE: A Text-guided Hierarchical Noise Eliminator for Multimodal Sentiment Analysis (2025.coling-main)
Copied to clipboard
| Challenge: | Existing methods for multimodal sentiment analysis assume that all modalities contribute equally to model performance. |
| Approach: | They propose a text-guided Hierarchical Noise Eliminator model that extracts modality-consistent information from unimodal data and integrates it into multimodal representations for sentiment classification. |
| Outcome: | The proposed model reduces noise caused by modality inconsistency by maximizing mutual information between textual representations and visual and acoustic representations. |
SPECTER: Document-level Representation Learning using Citation-informed Transformers (2020.acl-main)
Copied to clipboard
| Challenge: | Recent Transformer language models do not leverage information on inter-document relatedness, which limits their document-level representation power. |
| Approach: | They propose a method to generate document-level embeddings using citation graphs. |
| Outcome: | The proposed method outperforms baselines on document-level tasks. |
Assessing Multilingual Fairness in Pre-trained Multimodal Representations (2022.findings-acl)
Copied to clipboard
| Challenge: | Recent pre-trained multimodal models have shown exceptional capabilities towards connecting images and natural language. |
| Approach: | They propose two new fairness notions for pre-trained multimodal models that consider language as the fairness recipient. |
| Outcome: | The proposed models can be generalized to multilingualism by cross-lingual alignment . the results show that the models are individually fair across languages . |
UNIMO-2: End-to-End Unified Vision-Language Grounded Learning (2022.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for vision-language pre-training can only learn from aligned image-caption data and rely heavily on expensive regional features. |
| Approach: | They propose an end-to-end unified-modal pre-training framework for joint learning . they propose to conduct grounded learning on both images and texts via a sharing grounded space . |
| Outcome: | The proposed model improves visual and visual semantic alignment on images and texts. |
ProMedTS: A Self-Supervised, Prompt-Guided Multimodal Approach for Integrating Medical Text and Time Series (2025.findings-acl)
Copied to clipboard
Shuai Niu, Jing Ma, Hongzhan Lin, Liang Bai, Zhihua Wang, V. W., Richard Yi Da Xu, Guo Li, Xian Yang
| Challenge: | Large language models excel at processing unstructured data, but integrating time series data with text remains a challenge. |
| Approach: | They propose a self-supervised multimodal framework that uses prompt-guided learning to unify heterogeneous data types. |
| Outcome: | The proposed framework outperforms state-of-the-art approaches on disease diagnosis tasks using real-world datasets. |
SEEN: Structured Event Enhancement Network for Explainable Need Detection of Information Recall Assistance (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing work on information recall focuses on reactively retrieving life events . but, proactively detecting the need for information recall services is rarely discussed . |
| Approach: | They propose a human-annotated life experience retelling dataset to detect the right time to trigger an information recall service. |
| Outcome: | The proposed system detects life event inconsistency, additional information in life events, and forgotten events. |
The Integration of Semantic and Structural Knowledge in Knowledge Graph Entity Typing (2024.naacl-long)
Copied to clipboard
| Challenge: | Existing methods to predict missing type annotations for knowledge graphs use only structural knowledge in the local neighborhood of entities. |
| Approach: | They propose a model for KG Entity Typing that integrates semantic and structural knowledge to infer missing types. |
| Outcome: | The proposed framework outperforms existing state-of-the-art methods in the Knowledge Graph Entity Typing task. |
Unmasking the Hidden Meaning: Bridging Implicit and Explicit Hate Speech Embedding Representations (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods to detect explicit hate speech (HS) are focusing on detecting explicit forms of hateful expressions on user-generated content. |
| Approach: | They propose to examine the differences between embedding implicit and explicit hateful messages . they compare and link explicit and implicit hateful message across datasets . |
| Outcome: | The proposed model improves on explicit hate speech detection while retaining high performance on borderline cases. |
Interpreting Character Embeddings With Perceptual Representations: The Case of Shape, Sound, and Color (2022.acl-long)
Copied to clipboard
| Challenge: | Character-level information is included in many NLP models, but evaluating the information encoded in character embeddings is an open issue. |
| Approach: | They propose to use shape, sound, and color embeddings to evaluate the information encoded in character representations in five languages to perform cross-lingual analysis. |
| Outcome: | The proposed classifiers evaluate phonological information encoded in character embeddings and LSTM models. |
Zero-Shot Data Maps. Efficient Dataset Cartography Without Model Training (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods to diagnose large annotated datasets require the fitting of a strong model to the dataset. |
| Approach: | They propose a new approach to compute confidence and variability over an ensemble of zero-shot models constructed with different but semantically equivalent label descriptions. |
| Outcome: | The proposed method can be used to diagnose large annotated datasets with accuracy up to 14x faster than the current method. |
Learning Co-Speech Gesture for Multimodal Aphasia Type Detection (2023.emnlp-main)
Copied to clipboard
| Challenge: | Aphasia is a language disorder caused by brain damage affecting speech functions . a detailed diagnosis of aphasia type is imperative for effective treatment . but, little attention has been paid to developing methods to detect different types of sphasis . |
| Approach: | They propose a multimodal graph neural network for aphasia type detection using co-speech gestures and corresponding speech and gesture patterns. |
| Outcome: | The proposed model outperforms existing methods in F1 and 84.2% of cases. |
PAT: Parameter-Free Audio-Text Aligner to Boost Zero-Shot Audio Classification (2025.naacl-long)
Copied to clipboard
| Challenge: | Audio-Language Models (ALMs) have demonstrated remarkable performance in zero-shot audio classification. |
| Approach: | They propose a training-free method that enhances audio and language representations using mutual feedback. |
| Outcome: | The proposed method outperforms vanilla zero-shot evaluation with significant margins of 0.42%-27.0%. |
HRCA+: Advanced Multiple-choice Machine Reading Comprehension Method (2022.lrec-1)
Copied to clipboard
| Challenge: | Multiple-choice question answering (MCQA) requires a model to understand natural languages and understand textual representations. |
| Approach: | They propose a model that uses human reading comprehension attention to increase accuracy for machine reading comprehension. |
| Outcome: | The proposed model outperforms state-of-the-art models on the Semeval-2018 Task 11 dataset and on the DREAM dataset. |
LogogramNLP: Comparing Visual and Textual Representations of Ancient Logographic Writing Systems for NLP (2024.acl-long)
Copied to clipboard
| Challenge: | Existing pipelines for natural language processing only process symbolic representations of language, which are labor-intensive and noisy . a large portion of logographic data persists in a purely visual form due to the absence of transcription . this issue poses a bottleneck for researchers seeking to apply NLP to ancient logographic languages . |
| Approach: | They propose a benchmark for NLP analysis of ancient logographic languages using visual representations of writing. |
| Outcome: | The proposed pipeline outperforms existing pipelines for some tasks . the results could unlock large amounts of cultural heritage data of ancient logographic languages . |
Scaling Text-Rich Image Understanding via Code-Guided Synthetic Multimodal Data Generation (2025.acl-long)
Copied to clipboard
Yue Yang, Ajay Patel, Matt Deitke, Tanmay Gupta, Luca Weihs, Andrew Head, Mark Yatskar, Chris Callison-Burch, Ranjay Krishna, Aniruddha Kembhavi, Christopher Clark
| Challenge: | Vision-language models struggle to understand text-rich images due to the scarcity of diverse text-only large language data. |
| Approach: | They propose a framework that leverages the coding capabilities of text-only large language models to create synthetic text-rich multimodal data. |
| Outcome: | The proposed framework can generate high-quality instruction-tuning data using Python, HTML, LaTeX and other languages. |
UnitY: Two-pass Direct Speech-to-speech Translation with Discrete Units (2023.acl-long)
Copied to clipboard
Hirofumi Inaguma, Sravya Popuri, Ilia Kulikov, Peng-Jen Chen, Changhan Wang, Yu-An Chung, Yun Tang, Ann Lee, Shinji Watanabe, Juan Pino
| Challenge: | Experimental evaluations show that UnitY outperforms a single-pass speech-to-unit translation model by 2.5-4.2 ASR-BLEU with 2.83x decoding speed-up. |
| Approach: | They propose a two-pass direct S2ST architecture which generates textual representations and predicts discrete acoustic units . they show that UnitY outperforms a single-pass speech-to-unit translation model by 2.5-4.2 ASR-BLEU with 2.83x decoding speed-up. |
| Outcome: | The proposed architecture outperforms a single-pass speech-to-unit translation model by 2.5-4.2 ASR-BLEU with 2.83x decoding speed-up on large datasets. |
Composition-contrastive Learning for Sentence Embeddings (2023.acl-long)
Copied to clipboard
| Challenge: | Recent work shows potential to learn vector representations from unlabelled data without task-specific fine-tuning. |
| Approach: | They propose to maximize alignment between textual embeddings and a composition of their phrasal constituents. |
| Outcome: | The proposed approach improves on similarity tasks comparable to state-of-the-art approaches. |
See Detail Say Clear: Towards Brain CT Report Generation via Pathological Clue-driven Representation Learning (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Brain CT report generation is important to aid physicians in diagnosing cranial diseases. |
| Approach: | They propose a Pathological Clue-driven Representation Learning model to build cross-modal representations based on pathological clues and adapt them for text generation. |
| Outcome: | The proposed method outperforms previous methods and achieves SoTA performance. |
Vision-Language Models Struggle to Align Entities across Modalities (2025.findings-acl)
Copied to clipboard
| Challenge: | Several real-world applications require the ability to perform cross-modal entity linking . cross-functional entity linking is a skill needed for multimodal code generation and scene understanding . |
| Approach: | They propose a task and benchmark to evaluate cross-modal entity linking performance . they use visual scenes aligned with their textual representations to evaluate performance a question-answering task . |
| Outcome: | The proposed task and benchmark aims to improve cross-modal entity linking performance . it evaluates state-of-the-art vision-language models and humans on the task . |
MEVTR: A Multilingual Model Enhanced with Visual Text Representations (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing models that generate multilingual text representations perform poorly on low-resource languages due to lack of representation space and model capacity. |
| Approach: | They propose a multilingual model enhanced with visual text representations which complements textual representations and extends multilingual representation space with visual representations. |
| Outcome: | The proposed model outperforms state-of-the-art models on zero-shot cross-lingual transfer tasks without the target language adapter. |
Dialogue Act-Aided Backchannel Prediction Using Multi-Task Learning (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Backchanneling is a form of feedback that is produced by listeners in a conversation . since the advent of ChatGPT, modern dialogue systems exhibit answer quality levels on par with humans in various professions. |
| Approach: | They propose a multi-task learning approach that learns textual representations for the task of backchannel prediction in tandem with dialogue act classification. |
| Outcome: | The proposed approach improves the prediction of specific backchannels by up to 2.0% in F1 . the audio encoder is pre-trained in a self-supervised fashion using voice activity projection . |
Textual Steering Vectors Can Improve Visual Understanding in Multimodal Large Language Models (2026.acl-long)
Copied to clipboard
Woody Haosheng Gan, Deqing Fu, Julian Asilis, Ollie Liu, Vatsal Sharan, Robin Jia, Willie Neiswanger
| Challenge: | Steering methods have emerged as effective tools for guiding large language models’ behavior, yet multimodal large language model lacks comparable techniques due to architectural diversity and limited availability of multimodal steering vectors. |
| Approach: | They validate steering vectors derived solely from text-only LLM backbones and use a cross-modal transfer technique to reuse existing interpretability tools. |
| Outcome: | The proposed steering vectors can guide and enhance multimodal models using SPAR, Mean Shift, and Linear Probing. |
Modeling Human Perspectives with Socio-Demographic Representations (2026.findings-acl)
Copied to clipboard
| Challenge: | Recent studies show that human disagreement is widespread across many annotation tasks. |
| Approach: | They propose a method that jointly models annotator perspectives while learning socio-demographic representations. |
| Outcome: | The proposed method outperforms concatenation-based methods in predicting annotator perspectives . it learns socio-demographic representations and analyzes how demographic factors relate to variation . |