Papers by Houfeng Wang
Two Pathways to Truthfulness: On the Intrinsic Encoding of LLM Hallucinations (2026.acl-long)
Copied to clipboard
Wen Luo, Guangyue Peng, Wei Li, Shaohang Wei, Feifan Song, Liang Wang, Nan Yang, Xingxing Zhang, Jing Jin, Furu Wei, Houfeng Wang
| Challenge: | Previous work shows that large language models generate hallucinations, yet the origins and mechanisms of these signals remain unclear. |
| Approach: | They propose to validate and disentangle two different pathways for truthfulness cues . they also propose to use the same mechanism to derive self-contained evidence from the generated answer . |
| Outcome: | The proposed applications improve hallucination detection performance by integrating two different inputs. |
Encode Errors: Representational Retrieval of In-Context Demonstrations for Multilingual Grammatical Error Correction (2025.findings-acl)
Copied to clipboard
| Challenge: | a novel method for encoding fine-grained error patterns improves performance on GEC. |
| Approach: | They propose a method for encoding grammatical errors from LLMs' internal states using a GER method. |
| Outcome: | The proposed method significantly boosts performance in ICL settings on multilingual GEC datasets. |
Investigating the (De)Composition Capabilities of Large Language Models in Natural-to-Formal Language Conversion (2025.naacl-long)
Copied to clipboard
| Challenge: | Existing frameworks for evaluating the decomposition and composition capabilities of large language models (LLMs) in N2F are inadequate, and there are errors that can be attributed to deficiencies in natural language understanding and the learning and use of symbolic systems. |
| Approach: | They propose a framework that semi-automatically performs sample and task construction . main findings include that LLMs are deficient in both decomposition and composition . |
| Outcome: | The proposed framework evaluates the most advanced LLMs on a variety of common formal languages. |
Detection-Correction Structure via General Language Model for Grammatical Error Correction (2024.acl-long)
Copied to clipboard
| Challenge: | Grammatical error correction (GEC) is a task dedicated to rectifying texts with minimal edits. |
| Approach: | They propose a detection-correction structure based on the general language model which integrates detection and correction into a single model. |
| Outcome: | The proposed model outperforms the state-of-the-art models on English and Chinese datasets. |
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. |
MC2: A Minimum-Coverage and Dataset-Agnostic Framework for Compositional Generalization of LLMs on Semantic Parsing (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing research relies on dataset-specific designs or a large number of samples to improve compositional generalization of large language models (LLMs) . |
| Approach: | They propose a minimum-coverage framework that can help LLMs achieve compositional generalization by selecting and organizing samples that satisfy the primitive coverage. |
| Outcome: | The proposed framework can improve compositional generalization on different parsing datasets in the minimum-coverage setting. |
Auto-Dialabel: Labeling Dialogue Data with Unsupervised Learning (D18-1)
Copied to clipboard
| Challenge: | Existing dialog datasets rely on human labeling, which is expensive, limited in size, and in low coverage. |
| Approach: | They propose a framework to automatically cluster dialogue intents and slots . they collect context features, leverage an autoencoder for feature assembly, and adapt a dynamic hierarchical clustering method for intent and slot labeling. |
| Outcome: | The proposed framework can promote human labeling cost to a great extent and achieve good intent clustering accuracy (84.1%) it also provides reasonable and instructive slot labeling results. |
Adjusting the Precision-Recall Trade-Off with Align-and-Predict Decoding for Grammatical Error Correction (2022.acl-short)
Copied to clipboard
| Challenge: | Modern writing assistance applications always contain a Grammatical Error Correction (GEC) model to correct errors in user-entered sentences. |
| Approach: | They propose a simple yet effective approach to Align-and-Predict Decoding for most popular sequence-to-sequence models to offer more flexibility for the precision-recall trade-off. |
| Outcome: | The proposed model can be used in both English and Chinese GEC models and achieve state-of-the-art results. |
Investigating More Explainable and Partition-Free Compositionality Estimation for LLMs: A Rule-Generation Perspective (2026.acl-long)
Copied to clipboard
| Challenge: | Compositional generalization tests focus on output results without considering sample compositionality, resulting in explainability defects. |
| Approach: | They propose a rule-generation perspective for compositionality estimation for LLMs that requires LLM to generate a program as rules for dataset mapping and provides estimates of compositionality using complexity-based theory. |
| Outcome: | The proposed model provides estimates of the compositionality of LLMs using complexity-based theory on a string-to-grid task. |
HPT: Hierarchy-aware Prompt Tuning for Hierarchical Text Classification (2022.emnlp-main)
Copied to clipboard
| Challenge: | Hierarchical text classification (HTC) is a multi-label classification problem with a complex label hierarchy. |
| Approach: | They propose a Hierarchy-aware Prompt Tuning method to handle HTC from a multi-label perspective using a dynamic virtual template and label words that take the form of soft prompts to fuse the label hierarchy knowledge. |
| Outcome: | The proposed method achieves state-of-the-art performance on 3 popular HTC datasets and is adept at handling imbalance and low resource situations. |
M3: A Multi-View Fusion and Multi-Decoding Network for Multi-Document Reading Comprehension (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for multi-document reading comprehension cannot make full of the advantages of both approaches. |
| Approach: | They propose a multi-view fusion and multi-decoding method that integrates multiple documents for answering questions. |
| Outcome: | The proposed method improves on two mainstream multi-document reading comprehension datasets. |
Learning Invariant Representation Improves Robustness for MRC Models (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Existing approaches to improve machine reading comprehension models are vulnerable and not robust to adversarial examples. |
| Approach: | They propose to construct positive example pairs which have same answer by augmentation and then introduce stability and contrastive loss to improve invariance of representation. |
| Outcome: | The proposed approach boosts the robustness of QA models across different tasks and attack sets significantly and consistently. |
Odysseus Navigates the Sirens’ Song: Dynamic Focus Decoding for Factual and Diverse Open-Ended Text Generation (2025.acl-long)
Copied to clipboard
| Challenge: | Existing decoding methods struggle to balance factuality and diversity . Deterministic decoding approaches suffer from degeneration and lack of diversity - a problem that is not addressed by the current literature. |
| Approach: | They propose a plug-and-play stochastic approach that adjusts decoding focus based on distributional differences across layers, leveraging the modular nature of factual knowledge within LLMs. |
| Outcome: | Extensive experiments on seven datasets show that DFD significantly improves performance. |
Joint Learning for Targeted Sentiment Analysis (D18-1)
Copied to clipboard
| Challenge: | Recent studies have attempted to perform two tasks separately, e.g., target extraction and sentiment classification. |
| Approach: | They propose a hierarchical stack bidirectional gated recurrent units (HSBi-GRU) model which allows the target label to influence their sentiment label. |
| Outcome: | The proposed model outperforms baseline models on two datasets and shows that it can learn abstract features. |
Zero-shot Cross-lingual Transfer of Prompt-based Tuning with a Unified Multilingual Prompt (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing work focuses on monolingual prompts, but we study multilingual prompt for multilingual models. |
| Approach: | They propose a model that uses a unified prompt for all languages, called UniPrompt, to alleviate the effort of designing different prompts for multiple languages. |
| Outcome: | The proposed model outperforms baseline models in the zero-shot cross-lingual setting. |
simNet: Stepwise Image-Topic Merging Network for Generating Detailed and Comprehensive Image Captions (D18-1)
Copied to clipboard
| Challenge: | Existing approaches to image captioning combine visual and semantic attention to generate a detailed and comprehensive caption. |
| Approach: | They propose a stepwise image-topic merging network that integrates visual and semantic attentions to generate a detailed caption. |
| Outcome: | The proposed approach is evaluated on two benchmark datasets and reaches the state-of-the-art performance. |
Select High-quality Synthetic QA Pairs to Augment Training Data in MRC under the Reward Guidance of Generative Language Models (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing approaches focus on downstream metrics to select QA pairs, which lack generalization across different datasets. |
| Approach: | They propose a general selection method that uses a large pre-trained language model as a reward model in a Reinforcement Learning framework for the training of the selection agent. |
| Outcome: | The proposed method improves performance on generative and extractive datasets. |
Utilizing Local Hierarchy with Adversarial Training for Hierarchical Text Classification (2024.lrec-main)
Copied to clipboard
| Challenge: | Hierarchical text classification (HTC) is a challenging subtask due to its complex taxonomic structure. |
| Approach: | They propose a local hierarchy framework that can fit in nearly all HTC models and optimize them with the local hierarchy as auxiliary information. |
| Outcome: | The proposed framework is effective in all scenarios and is adept at dealing with complex taxonomic hierarchies. |
Exploring Sequence-to-Sequence Learning in Aspect Term Extraction (P19-1)
Copied to clipboard
| Challenge: | Aspect term extraction (ATE) aims at identifying all aspect terms in a sentence . sequence labeling based methods cannot make full use of overall meaning of sentence if they have dependencies between labels. |
| Approach: | They propose to formalize ATE as a sequence-to-sequence (Seq2Seque) learning task . they propose gated unit networks and position-aware attention mechanism to make it suit to ATE . |
| Outcome: | The proposed learning task is effective when labels correspond to words one by one . the proposed learning system is gated unit networks and position-aware attention mechanism . |
Original Content Is All You Need! an Empirical Study on Leveraging Answer Summary for WikiHowQA Answer Selection Task (2022.coling-1)
Copied to clipboard
Liang Wen, Juan Li, Houfeng Wang, Yingwei Luo, Xiaolin Wang, Xiaodong Zhang, Zhicong Cheng, Dawei Yin
| Challenge: | Existing answer selection approaches for community question answering lack additional answer summaries due to redundancy and lengthiness issues of crowdsourced answers. |
| Approach: | They constructed a dataset which contains a corresponding reference summary for each original lengthy answer. |
| Outcome: | The proposed model improves the performance of a question and candidate answer on a WikiHowQA dataset. |
Instantaneous Grammatical Error Correction with Shallow Aggressive Decoding (2021.acl-long)
Copied to clipboard
| Challenge: | Existing approaches to improve online inference efficiency of the Transformer for instantaneous Grammatical Error Correction (GEC) are sequenceto-sequence (seq2sequ) and sequenceto sequence (saq2eq) |
| Approach: | They propose a novel approach to improve the online inference efficiency of the Transformer model for instantaneous Grammatical Error Correction (GEC) it aggressively decodes as many tokens as possible in parallel instead of always decoding only one token in each step to improve computational parallelism. |
| Outcome: | The proposed approach can achieve state-of-the-art results in English and Chinese benchmarks with 10x speedup over the Transformer-big model. |
Syntax-Aware Graph Attention Network for Aspect-Level Sentiment Classification (2020.coling-main)
Copied to clipboard
| Challenge: | Existing approaches to aspect-level sentiment classification focus on modeling the relationship between aspect words and their contexts with attention, and ignore the use of elaborate knowledge implicit in the context. |
| Approach: | They exploit syntactic awareness to the model by the graph attention network on the dependency tree structure and external pre-training knowledge by BERT language model, which helps to model the interaction between the context and aspect words better. |
| Outcome: | The proposed model can model the interaction between the context and aspect words better by using syntactic awareness and external pre-training knowledge. |
Attention Weights as an Indicator: Analyzing and Improving Document Utilization in Retrieval-Augmented Generation (2026.acl-long)
Copied to clipboard
| Challenge: | In traditional RAG models, documents are grouped into categories based on their quality and order, and the quality of inputs is variable due to ineffective retrievers or misalignment between the retriever and generator. |
| Approach: | They propose to use attention weights to enhance document utilization from three perspectives: document ranking, placement, and filtering. |
| Outcome: | The proposed method outperforms baselines and improves document utilization effectiveness in a training-free manner. |
SGM: Sequence Generation Model for Multi-label Classification (C18-1)
Copied to clipboard
| Challenge: | Existing methods ignore the correlations between labels and different parts of the text can contribute differently for predicting different labels. |
| Approach: | They propose to view the multi-label classification task as a sequence generation problem and apply a decoder-based sequence generation model to solve it. |
| Outcome: | The proposed methods outperform previous work by a substantial margin. |
FEA-Bench: A Benchmark for Evaluating Repository-Level Code Generation for Feature Implementation (2025.acl-long)
Copied to clipboard
Wei Li, Xin Zhang, Zhongxin Guo, Shaoguang Mao, Wen Luo, Guangyue Peng, Yangyu Huang, Houfeng Wang, Scarlett Li
| Challenge: | Existing benchmarks focus on standalone programming problems, such as HumanEval, MBPP, and LiveCodeBench. |
| Approach: | They propose to use large language models to evaluate their ability to perform incremental development within code repositories by collecting pull requests from 83 GitHub repositorias and using rule-based and intent-based filtering to construct task instances focused on new feature development. |
| Outcome: | The proposed benchmarks show that large language models perform significantly worse in the FEA-Bench, highlighting considerable challenges in repository-level incremental code development. |
Learn to Memorize: Scalable Continual Learning in Semiparametric Models with Mixture-of-Neighbors Induction Memory (2025.acl-long)
Copied to clipboard
| Challenge: | Semiparametric language models (LMs) use static storage, which lacks learning capability and is disconnected from the internal information flow of the parametric models. |
| Approach: | They reconceptualize the non-parametric memory represented by kNN-LM as a learnable Mixture-of-Neighbors Induction Memory (MoNIM) this synergizes the induction capabilities of attention heads with the memorization strength of feed-forward networks . |
| Outcome: | The proposed model is a learnable Mixture-of-neighbors induction memory (MoNIM) it synergizes the induction capabilities of attention heads with the memorization strength of feed-forward networks (FFNs). |
Beyond Transcription: Unified Audio Schema for Perception-Aware AudioLLMs (2026.findings-acl)
Copied to clipboard
Linhao Zhang, Yuhan Song, Aiwei Liu, Chuhan Wu, Sijun Zhang, Wei Jia, Yuan Liu, Houfeng Wang, Zhou Xiao
| Challenge: | Recent Audio Large Language Models (AudioLLMs) excel at reasoning tasks, but struggle at elementary auditory perception. |
| Approach: | They propose a framework that organizes audio information into three explicit components in a unified JSON format. |
| Outcome: | The proposed framework boosts fine-grained perception by 10.9% on MMSU over state-of-the-art models while preserving robust reasoning capabilities. |
Question Condensing Networks for Answer Selection in Community Question Answering (P18-1)
Copied to clipboard
| Challenge: | Community question answering (CQA) is a subtask of community question answering . previous researches ignored the difference between the two parts and concatenated them as the question representation . |
| Approach: | They propose a question condensing network that makes use of the subject-body relationship of community questions. |
| Outcome: | The proposed model outperforms existing models on two CQA datasets. |
SPOR: A Comprehensive and Practical Evaluation Method for Compositional Generalization in Data-to-Text Generation (2024.acl-long)
Copied to clipboard
| Challenge: | Existing studies on compositional generalization in data-to-text generation focus on one manifestation, Systematicity, Productivity, Order invariance, and Rule learnability. |
| Approach: | They propose a method for evaluation of compositional generalization in data-to-text generation that includes four aspects of manifestations and allows high-quality evaluation without additional manual annotations. |
| Outcome: | The proposed method is based on two datasets and evaluates existing language models including LLMs. |
Scaling Data Diversity for Fine-Tuning Language Models in Human Alignment (2024.lrec-main)
Copied to clipboard
| Challenge: | Large language models (LLMs) can reveal toxic or offensive content inadvertently or intentionally. |
| Approach: | They propose to control the diversity of both sides according to the number of samples for fine-tuning, which can directly reflect their impact. |
| Outcome: | The proposed approach improves the performance of large language models after fine-tuning. |
Would You Like to Make a Donation? A Dialogue System to Persuade You to Donate (2024.lrec-main)
Copied to clipboard
| Challenge: | Persuasive automated dialogue systems are a popular way to influence people's behavior and decision making. |
| Approach: | They propose to use a context-aware persuasion strategy selection module to persult users . they also propose a persuasiveness prediction model to automatically evaluate the persuasiveness of generated text. |
| Outcome: | The proposed system can achieve better performance on several automated evaluation metrics than baseline models. |
A Neural Question Answering Model Based on Semi-Structured Tables (C18-1)
Copied to clipboard
| Challenge: | Existing question answering systems rely on raw text and structured knowledge graphs. |
| Approach: | They build an end-to-end system to answer multiple choice questions with semi-structured tables as its knowledge. |
| Outcome: | The proposed system improves on the state-of-the-art question answering system with tabMCQ dataset. |
Do It Once: An Embarrassingly Simple Joint Matching Approach to Response Selection (2021.findings-acl)
Copied to clipboard
| Challenge: | Existing matching models for response selection perform the independent matching (IM) approach. Existing models for matching only perform one match regardless of the number of options. |
| Approach: | They propose a joint matching approach which performs matching only once regardless of the number of options. |
| Outcome: | The proposed approach outperforms existing models and reduces training time by over half. |
Multi-Layer Pseudo-Siamese Biaffine Model for Dependency Parsing (2022.coling-1)
Copied to clipboard
| Challenge: | Existing work only uses biaffine method at the end of the dependency parser as a scorer, and its application in multi-layer form is ignored. |
| Approach: | They propose a multi-layer pseudo-Siamese biaffine model for neural dependency parsing that uses biaffin method as a scorer and a biaffin module to construct arc weight matrix. |
| Outcome: | The proposed model achieves state-of-the-art on PTB, CTB, and UD datasets with low efficiency loss. |
Well Begun is Half Done: Low-resource Preference Alignment by Weak-to-Strong Decoding (2025.findings-acl)
Copied to clipboard
| Challenge: | Low-resource methods for LLM alignment have been popular, but still face challenges in obtaining high-quality and aligned content. |
| Approach: | They propose a framework to enhance alignment ability of base models by the guidance of a small aligned model. |
| Outcome: | The proposed framework outperforms baseline methods while avoiding degradation on downstream tasks. |
DVD: Dynamic Contrastive Decoding for Knowledge Amplification in Multi-Document Question Answering (2024.emnlp-main)
Copied to clipboard
| Challenge: | Large language models (LLMs) generate information with hallucinations due to uneven retrieval quality and irrelevant contents. |
| Approach: | They propose a decoding strategy which dynamically amplifies knowledge from selected documents during the generation phase. |
| Outcome: | The proposed method outperforms other decoding strategies on ALCE-ASQA, NQ, TQA and PopQA benchmarks. |
Incorporating Hierarchy into Text Encoder: a Contrastive Learning Approach for Hierarchical Text Classification (2022.acl-long)
Copied to clipboard
| Challenge: | Existing methods encode text and label hierarchy separately and mix their representations for classification, where the hierarchy remains unchanged for all input text. |
| Approach: | They propose to embed hierarchy into a text encoder by combining input and output data to generate a hierarchy-aware representation. |
| Outcome: | Extensive experiments on three benchmark datasets verify the effectiveness of the proposed model. |
Text Level Graph Neural Network for Text Classification (D19-1)
Copied to clipboard
| Challenge: | Recent researches have explored graph neural network (GNN) techniques on text classification, but they are faced with the problems of fixed corpus level graph structure which don’t support online testing and high memory consumption. |
| Approach: | They propose a graph neural network model that builds graphs for each input text with global parameters sharing instead of a single graph for the whole corpus. |
| Outcome: | The proposed model outperforms existing models on several text classification datasets even with consuming less memory. |
Instantly Learning Preference Alignment via In-context DPO (2025.naacl-long)
Copied to clipboard
| Challenge: | Presently, mainstream approaches to HPA heavily depend on fine-tuning . however, the huge computational and annotation costs of fine-timing are hard to ignore . |
| Approach: | They propose a tuning-free approach to HPA using LLMs' decoding . they first rethink the derivation procedures of DPO and build an instant scorer . |
| Outcome: | The proposed approach outperforms existing methods even with tuning-free baselines and an upgraded scorer. |
Unpaired Sentiment-to-Sentiment Translation: A Cycled Reinforcement Learning Approach (P18-1)
Copied to clipboard
| Challenge: | Existing studies for sentiment-to-sentiment "translation" only change the underlying sentiment and fail to keep the semantic content. |
| Approach: | They propose a cycled reinforcement learning method that combines neutralization module and emotionalization module. |
| Outcome: | The proposed method outperforms state-of-the-art systems on Yelp and Amazon review datasets. |
Autoencoder as Assistant Supervisor: Improving Text Representation for Chinese Social Media Text Summarization (P18-2)
Copied to clipboard
| Challenge: | Existing abstractive text summarization models learn a semantic representation of the source text and the summaries from it. |
| Approach: | They evaluate the model on a popular Chinese social media dataset and compare it to other models. |
| Outcome: | The proposed model achieves state-of-the-art performance on a popular Chinese social media dataset. |
Phrase-level Self-Attention Networks for Universal Sentence Encoding (D18-1)
Copied to clipboard
| Challenge: | Phrase-level self-attention networks (PSAN) can capture context dependencies at the phrase level instead of the sentence level. |
| Approach: | They propose to perform self-attention across words inside a phrase to capture context dependencies at the phrase level and use the gated memory updating mechanism to refine each word’s representation hierarchically with longer-term context dependency captured in a larger phrase. |
| Outcome: | The proposed model can achieve state-of-the-art performance across a plethora of NLP tasks including binary and multi-class classification, natural language inference and sentence similarity. |