Papers by Xiaojun Wan

CE-RM: A Pointwise Generative Reward Model Optimized via Two-Stage Rollout and Unified Criteria (2026.findings-acl)

Challenge:	Existing models on context-dependent text-to-SQL task focus on utilizing historic user inputs.
Approach:	They propose a database schema interaction graph encoder to utilize historic user inputs.
Outcome:	The proposed model outperforms previous state-of-the-art models on two datasets . it also outperformed existing models on the benchmark SParC and CoSQL datasets.

Copied to clipboard

Xinyu Hu, Yancheng He, Weixun Wang, Tao Feng, Li Lin, Jiashun Liu, Wenbo Su, Bo Zheng, Xiaojun Wan

Challenge:	Existing studies have shown that rule-based evaluation methods are ineffective for open-ended natural language generation.
Approach:	They propose a pointwise generative reward model with a dedicated two-stage rollout method and unified query-based criteria that can be trained with 5.7K high-quality data.
Outcome:	The proposed model achieves superior performance on diverse reward model benchmarks, especially in Best-of-N scenarios, and delivers more effective improvements in downstream RL practice.

Learning Bilingual Sentiment-Specific Word Embeddings without Cross-lingual Supervision (N19-1)

Copied to clipboard

Yanlin Feng, Xiaojun Wan

Multilinguality

Challenge:	Unsupervised BWE methods are evaluated on word translation or word similarity tasks.
Approach:	They propose a method that learns sentiment-specific word representations for two languages in a common space without cross-lingual supervision.
Outcome:	The proposed method outperforms previous unsupervised BWE methods and even supervised Bwe methods on three language pairs for cross-lingual sentiment analysis.

Gödel Agent: A Self-Referential Agent Framework for Recursively Self-Improvement (2025.acl-long)

Copied to clipboard

Xunjian Yin, Xinyi Wang, Liangming Pan, Li Lin, Xiaojun Wan, William Yang Wang

From TDMA to CDMA: A Multi-bit Watermark for Diffusion Language Models (2026.acl-long)

Challenge:	Existing agentic systems cannot search the whole design space due to the restriction of human-designed components.
Approach:	They propose a Gödel Agent framework that allows agents to recursively improve themselves without relying on fixed algorithms or fixed algorithms.
Outcome:	The proposed framework surpasses manual crafted agents in performance, efficiency, and generalizability.

Copied to clipboard

Baizhou Huang, Xiaojun Wan

Reference Matters: Benchmarking Factual Error Correction for Dialogue Summarization with Fine-grained Evaluation Framework (2023.acl-long)

Challenge:	Existing multi-bit watermarking schemes cannot be directly applied to DLMs.
Approach:	They propose a multi-bit watermarking framework that encodes the entire watermark message holographically.
Outcome:	The proposed framework encodes the entire watermark message across all tokens holographically.

Copied to clipboard

Mingqi Gao, Xiaojun Wan, Jia Su, Zhefeng Wang, Baoxing Huai

Knowledge Verification to Nip Hallucination in the Bud (2024.emnlp-main)

Challenge:	Current evaluations of FEC models that depend on factuality metrics are not reliable and detailed enough.
Approach:	They propose a fine-grained evaluation framework that automatically evaluates FEC models on different error categories.
Outcome:	The proposed evaluation framework compares models on different error categories and finds the best training modes and significant differences in the performance of existing models.

Copied to clipboard

Fanqi Wan, Xinting Huang, Leyang Cui, Xiaojun Quan, Wei Bi, Shuming Shi

Reasoning

Challenge:	Recent studies have shown that large language models generate responses that sound plausible but contradict factual knowledge, a phenomenon known as hallucination.
Approach:	They propose a novel approach to align large language models to evaluate knowledge boundaries based on external knowledge to reduce hallucinations .
Outcome:	The proposed approach reduces hallucinations across six benchmarks using foundation LLMs of varying backbones and scales.

CodeQA: A Question Answering Dataset for Source Code Comprehension (2021.findings-emnlp)

Copied to clipboard

Chenxiao Liu, Xiaojun Wan

Question Answering

Challenge:	False. a free-form question answering dataset can serve as a useful research benchmark for source code comprehension.
Approach:	They propose a free-form question answering dataset for source code comprehension . they implement syntactic rules and semantic analysis to transform code comments into question-answer pairs.
Outcome:	The proposed dataset can serve as a useful research benchmark for source code comprehension.

Towards A “Novel” Benchmark: Evaluating Literary Fiction with Large Language Models (2025.findings-acl)

Copied to clipboard

Wenqing Wang, Mingqi Gao, Xinyu Hu, Xiaojun Wan

Themis: A Reference-free NLG Evaluation Language Model with Flexibility and Interpretability (2024.emnlp-main)

Challenge:	Recent advances in Large Language Models (LLMs) context windows have enabled them to process inputs over 100K tokens and generate outputs of up to 10K token.
Approach:	They propose a multi-level evaluation framework that incorporates ten metrics across the Macro, Meso, and Micro levels and an annotated fiction dataset.
Outcome:	The proposed framework incorporates ten metrics across the Macro, Meso, and Micro levels and is based on a human-human-AI dataset.

Copied to clipboard

Xinyu Hu, Li Lin, Mingqi Gao, Xunjian Yin, Xiaojun Wan

UMMF: Protecting Copyright of Large Vision-Language Models through Unlearning-based Multimodal Memorization Fingerprint (2026.acl-long)

Challenge:	Existing methods for evaluation of natural language generation tasks lack reliable data.
Approach:	They propose to use annotations from human and GPT-4 to construct a corpus for NLG evaluation.
Outcome:	The proposed corpus can perform flexible and interpretable evaluations without references and surpasses existing models.

Copied to clipboard

Xiaofan Zheng, Xinghao Wang, Xiaojun Wan

Revisiting Pivot-Based Paraphrase Generation: Language Is Not the Only Optional Pivot (2021.emnlp-main)

Challenge:	Existing methods for fingerprinting large vision-Language Models rely on explicit triggers, which have limitations in terms of stealthiness and robustness.
Approach:	They propose to use model fingerprints to verify the ownership of large vision-Language Models (LVLMs) they use implicit model fingerprinting techniques that leverage neighboring samples as implicit model .
Outcome:	The proposed fingerprinting technique is superior to existing methods, but has limitations in terms of stealthiness and robustness.

Copied to clipboard

Yitao Cai, Yue Cao, Xiaojun Wan

Edit-Aware Reward Modeling for Chinese Grammatical Error Correction (2026.acl-long)

Challenge:	Existing methods for paraphrase generation rely on language as the pivot . however, there is no evidence that parallel data of paraphrases is needed for paraphrasing.
Approach:	They propose to use semantic and syntactic representations as pivot for paraphrase generation.
Outcome:	The proposed method can generate paraphrases with better quality than using language as pivot.

Copied to clipboard

Yilin Li, Xiaojun Wan

Image Matters: A New Dataset and Empirical Study for Multimodal Hyperbole Detection (2024.lrec-main)

Challenge:	Recent work has applied reinforcement learning with rule-based rewards to grammatical error correction tasks, but these methods fail to capture fine-grained quality distinctions among correction candidates.
Approach:	They propose an Edit-Aware Reward Model that explicitly incorporates edit-awareness into preference learning for CGEC.
Outcome:	The proposed model outperforms rule-based models on CGEC and other NLP tasks by 5.41 and 1.80 points.

Copied to clipboard

Huixuan Zhang, Xiaojun Wan

Multi-Modal Sarcasm Detection in Twitter with Hierarchical Fusion Model (P19-1)

Challenge:	linguistic detection of hyperbole is an important part of understanding human expression . studies on hyperbolic expressions focus on text modality, but social media can be used to detect it .
Approach:	They propose to use a multimodal detection dataset to study hyperbole detection . they treat text and image as two modalities and evaluate pre-trained encoders .
Outcome:	The proposed dataset is constructed from five different keywords and shows its performance.

Copied to clipboard

Yitao Cai, Huiyu Cai, Xiaojun Wan

INS: An Interactive Chinese News Synthesis System (N19-4)

Challenge:	Existing methods to detect sarcasm focus on text, but they are insufficient for multi-modal messages.
Approach:	They propose a multi-modal hierarchical sarcasm detection model for tweets consisting of texts and images in Twitter.
Outcome:	The proposed model is able to detect sarcasm on twitter using three modalities . the proposed model can be used in customer service, opinion mining and harassment detection .

Copied to clipboard

Hui Liu, Wentao Qin, Xiaojun Wan

Challenge:	In the last decade, news websites and apps become more popular, which can provide us an extremely large volume of news articles.
Approach:	They propose a system which automatically synthesizes news articles into a long overview article by interacting with users.
Outcome:	The proposed system can generate news overview articles automatically or by interacting with users.

Exploring Context-Aware Evaluation Metrics for Machine Translation (2023.findings-emnlp)

Copied to clipboard

Xinyu Hu, Xunjian Yin, Xiaojun Wan

Machine Translation

Challenge:	Existing studies on machine translation evaluation focused on quality of individual sentences, while neglecting the importance of contextual information.
Approach:	They propose a context-aware machine translation evaluation metric called Cont-COMET . they use the COMET framework to consider the preceding and subsequent contexts of the sentence .
Outcome:	The proposed metric improves system-level and segment-level evaluations on the official WMT framework.

Nearest Neighbor Knowledge Distillation for Neural Machine Translation (2022.naacl-main)

Copied to clipboard

Zhixian Yang, Renliang Sun, Xiaojun Wan

Exploring the Impact of Vision Features in News Image Captioning (2023.findings-acl)

Challenge:	k-nearest-neighbor machine translation (kNN-MT) is a state-of-the-art machine translation technique . however, it requires conducting kNN searches for each decoding step, which increases the cost of decoding .
Approach:	They propose to move the time-consuming kNN search forward to the preprocessing phase and introduce k Nearest Neighbor Knowledge Distillation (kNN-KD) that trains the base NMT model to directly learn the knowledge of kN.
Outcome:	The proposed method improves over the state-of-the-art model while maintaining the same training and decoding speed as the standard model.

Copied to clipboard

Junzhe Zhang, Xiaojun Wan

Are LLM-based Evaluators Confusing NLG Quality Criteria? (2024.acl-long)

Challenge:	Recent state-of-art models can achieve competitive performance even without vision features.
Approach:	They conduct extensive experiments with mainstream news image captioning models to determine whether vision features contribute to the generation of captions.
Outcome:	The proposed models can achieve competitive performance even without vision features.

Copied to clipboard

Xinyu Hu, Mingqi Gao, Sen Hu, Yang Zhang, Yicheng Chen, Teng Xu, Xiaojun Wan

Text Classification

Challenge:	Existing studies show that LLMs confuse evaluation criteria, which reduces their reliability.
Approach:	They propose a hierarchical classification system for 11 common aspects with corresponding different evaluation criteria.
Outcome:	The proposed system is based on 11 common aspects with different evaluation criteria.

Multi-Granularity Interaction Network for Extractive and Abstractive Multi-Document Summarization (2020.acl-main)

Copied to clipboard

Hanqi Jin, Tianming Wang, Xiaojun Wan

Information Extraction & Text Mining

Challenge:	Existing methods for document summarization use extractive and abstractive representations, but they don't take into account hierarchical structure of document clusters.
Approach:	They propose a multi-granularity interaction network for extractive and abstractive multi-document summarization which jointly learn semantic representations for words, sentences, and documents.
Outcome:	The proposed model outperforms baseline methods and achieves the best results on the Multi-News dataset.

New Datasets and Controllable Iterative Data Augmentation Method for Code-switching ASR Error Correction (2023.findings-emnlp)

Copied to clipboard

Zhaohong Wan, Xiaojun Wan, Wei Peng, Rongjun Li

Mutual-Taught for Co-adapting Policy and Reward Models (2025.acl-long)

Challenge:	In bilingual or multilingual settings, code-switching ASR has greater challenges and research value.
Approach:	They propose a controllable iterative method for improving the performance of mainstream automatic speech recognition systems by using Chinese-English code-switching dialogues.
Outcome:	The proposed method achieves the best performance compared with the rule-based, back-translation-based data augmentation methods and large language model ChatGPT.

Copied to clipboard

Tianyuan Shi, Canbin Huang, Fanqi Wan, Longguang Zhong, Ziyi Yang, Weizhou Shen, Xiaojun Quan, Ming Yan

Machine Translation

Challenge:	Experimental results show that this iterative approach leads to consistent improvements in both the policy model and reward model.
Approach:	They propose a method that iteratively improves both the policy model and reward model without requiring additional human annotation.
Outcome:	The proposed method improves both the policy model and reward model without human annotation.

MC-MKE: A Fine-Grained Multimodal Knowledge Editing Benchmark Emphasizing Modality Consistency (2025.findings-acl)

Copied to clipboard

Junzhe Zhang, Huixuan Zhang, Xunjian Yin, Baizhou Huang, Xu Zhang, Xinyu Hu, Xiaojun Wan

Ghost in the Shell: Synonym-Aware Logit Shaping Fingerprint for Copyright Protection of Large Vision-Language Models (2026.findings-acl)

Challenge:	Existing benchmarks for knowledge editing in multimodal large language models focus on limited scenarios due to the lack of rigorous definition of multimodal knowledge.
Approach:	They propose a decomposed definition of multimodal knowledge and a benchmark to evaluate it.
Outcome:	The proposed method reveals that it is difficult to define multimodal knowledge editing in LLMs.

Copied to clipboard

Xiaofan Zheng, Xinghao Wang, Xiaojun Wan

Abstractive Multi-Document Summarization via Joint Learning with Single-Document Summarization (2020.findings-emnlp)

Challenge:	Existing fingerprinting methods for large vision-language models rely on backdoors to elicit abnormal outputs, but direct distortion of the model’s original outputs compromises modality alignment and degrades multimodal capabilities.
Approach:	They propose to embed a robust fingerprint while preserving the original normal outputs of the model.
Outcome:	The proposed fingerprint maintains multimodal performance and substantially enhances fingerprint robustness.

Copied to clipboard

Hanqi Jin, Xiaojun Wan

Information Extraction & Text Mining

Challenge:	Existing methods for document summarization are extractive and abstractive.
Approach:	They propose to jointly learn an abstractive single-document decoder and a decoding controller to aggregate the decoded outputs for multiple input documents.
Outcome:	The proposed model outperforms several baselines on two multi-document summarization datasets and proves that it is useful for both tasks.

Is Summary Useful or Not? An Extrinsic Human Evaluation of Text Summaries on Downstream Tasks (2024.lrec-main)

Copied to clipboard

Xiao Pu, Mingqi Gao, Xiaojun Wan

Information Extraction & Text Mining

Challenge:	a recent study focused on intrinsic evaluation, which assesses the quality of summaries, e.g. coherence, fluency, and informativeness, but it focused on task-based extrinsic evaluation to determine the usefulness of summarizations.
Approach:	They incorporate three downstream tasks to measure the usefulness of summaries . they find that fine-tuned models produce more useful summary across all three tasks .
Outcome:	The proposed model produces more useful summaries across all three tasks compared to zero-shot models . human evaluation provides more reliable performance assessment compared with automatic methods .

RST-Guarder: Enhancing Long-Context Robustness for Safeguards via RST Parsing and Probabilistic Inference (2026.acl-long)

Copied to clipboard

Xu Zhang, Xiaojun Wan

Jointly Learning to Align and Summarize for Neural Cross-Lingual Summarization (2020.acl-main)

Challenge:	Existing guardrail models for harmful-content detection degrade on long-form inputs . Existing models are vulnerable to policy-violating responses, causing false positives based on benign content .
Approach:	They propose an inference-time method that improves harmful-content detection for long-form inputs without additional data curation or model training.
Outcome:	The proposed method improves harmful-content detection for long-form inputs without additional data curation or model training.

Copied to clipboard

Yue Cao, Hui Liu, Xiaojun Wan

Adversarial Text Generation via Sequence Contrast Discrimination (2020.findings-emnlp)

Challenge:	Existing studies on cross-lingual summarization focus on pipeline methods and training end-to-end models.
Approach:	They propose to jointly learn to align and align to train a neural cross-lingual summarization model by using a large-scale corpus.
Outcome:	The proposed model outperforms competing models in most cases and can generate cross-lingual summaries without access to any cross-linguistic corpus.

Copied to clipboard

Ke Wang, Xiaojun Wan

A Dual-Perspective NLG Meta-Evaluation Framework with Automatic Benchmark and Better Interpretability (2025.acl-long)

Challenge:	Existing approaches to generate human-like texts are auto-regressive, but they suffer from exposure bias due to the dependence on the previous sampled output during the inferring phase.
Approach:	They propose a sequence contrast loss driven text generation framework which learns the difference between real texts and generated texts and uses that difference.
Outcome:	The proposed framework improves training stability and quality of generated texts and avoids the time-consuming sampling process.

Copied to clipboard

Xinyu Hu, Mingqi Gao, Li Lin, Zhenghan Yu, Xiaojun Wan

Explainability & Interpretability in NLP

Challenge:	Existing evaluation metrics are insufficient to meet requirements for natural language generation.
Approach:	They propose a dual-perspective NLG meta-evaluation framework that focuses on different evaluation capabilities and a method of automatically constructing benchmarks without requiring new human annotations.
Outcome:	The proposed framework improves interpretability and provides better performance for 16 representative LLMs.

Minos: A Multimodal Evaluation Model for Bidirectional Generation Between Image and Text (2026.findings-acl)

Copied to clipboard

Junzhe Zhang, Huixuan Zhang, Xinyu Hu, Li Lin, Mingqi Gao, Shi Qiu, Xiaojun Wan

Evaluating Factuality in Cross-lingual Summarization (2023.findings-acl)

Challenge:	Existing evaluation models struggle to achieve consistent performance across image-to-text (I2T) and text-to image (T2I) tasks.
Approach:	They construct a multimodal evaluation model using a large multimodal dataset and rigorous quality control strategies to train it.
Outcome:	The proposed model achieves state-of-the-art evaluation performance across 16 out-of domain datasets covering both I2T and T2I tasks among all open-source multimodal evaluation models and remain competitive with closed-source models.

Copied to clipboard

Mingqi Gao, Wenqing Wang, Xiaojun Wan, Yuemei Xu

Clustering-Aware Negative Sampling for Unsupervised Sentence Representation (2023.findings-acl)

Challenge:	Existing evaluation metrics for monolingual summarization require translation to evaluate the factuality of cross-lingual summmarization.
Approach:	They propose to analyze cross-lingual factuality by collecting annotations and generated summaries from models at summary level and sentence level.
Outcome:	The proposed dataset shows that over 50% of generated summaries contain factual errors with different characteristics from monolingual summarization.

Copied to clipboard

Jinghao Deng, Fanqi Wan, Tao Yang, Xiaojun Quan, Rui Wang

Sense-Aware Neural Models for Pun Location in Texts (P18-2)

Challenge:	Using clustering-aware learning, in-batch negatives are often ignored in sentence representation learning.
Approach:	They propose a method that integrates cluster information into contrastive learning for unsupervised sentence representation learning.
Outcome:	The proposed method compares favorably with baselines on semantic textual similarity tasks.

Copied to clipboard

Yitao Cai, Yin Li, Xiaojun Wan

Visual Information Guided Zero-Shot Paraphrase Generation (2022.coling-1)

Challenge:	Puns where the two meanings share the same pronunciation are known as homographic puns.
Approach:	They propose a sense-aware neural model to address the task of pun location . they first obtain several WSD results for the text and then leverage a bidirectional LSTM network to model each word senses.
Outcome:	The proposed model is based on a SemEval 2017 benchmark dataset showing that it can predict homographic puns.

Copied to clipboard

Zhe Lin, Xiaojun Wan

ALCUNA: Large Language Models Meet New Knowledge (2023.emnlp-main)

Challenge:	Several studies use different information as ”pivot” such as language, semantic representation and so on.
Approach:	They propose to use visual information as the "pivot" of back-translation to generate paraphrases using paired image-caption data.
Outcome:	The proposed model generates paraphrase with good relevancy, fluency and diversity . it is based on paired image-caption data and can train a paraphrasing model .

Copied to clipboard

Xunjian Yin, Baizhou Huang, Xiaojun Wan

Style-Compress: An LLM-Based Prompt Compression Framework Considering Task-Specific Styles (2024.findings-emnlp)

Challenge:	Existing benchmarks do not adequately measure large-scale language models’ capabilities when faced with new knowledge.
Approach:	They propose a benchmark called ALCUNA to evaluate LLMs' ability to handle new knowledge by altering existing entity attributes and relationships.
Outcome:	The proposed approach generates new knowledge by altering existing entity attributes and relationships, resulting in artificial entities distinct from real-world entities.

Copied to clipboard

Xiao Pu, Tianxing He, Xiaojun Wan

Multimodal Transformer for Multimodal Machine Translation (2020.acl-main)

Challenge:	Prompt compression reduces inference time and costs while maintaining informativeness for different usage scenarios.
Approach:	They propose a framework that adapts a smaller language model to compress prompts for a larger model on a new task without additional training.
Outcome:	The proposed framework outperforms two baseline models in four tasks . iteratively generates and selects effective compressed prompts as task-specific demonstrations .

Copied to clipboard

Shaowei Yao, Xiaojun Wan

Multilinguality

Challenge:	Existing methods to incorporate information from other modality, usually static images, are not considered relative to multimodal machine translation.
Approach:	They propose a multimodal self-attention method which learns the representation of images based on the text, which avoids encoding irrelevant information in images.
Outcome:	The proposed model outperforms previous studies and competitive baselines in terms of various metrics.

Pushing Paraphrase Away from Original Sentence: A Multi-Round Paraphrase Generation Approach (2021.findings-acl)

Copied to clipboard

Zhe Lin, Xiaojun Wan

Self-Evolution Fine-Tuning for Policy Optimization (2024.findings-emnlp)

Challenge:	Recent years, neural paraphrase generation models have demonstrated superior performance, but the output paraphrase still lacks diversity.
Approach:	They propose a back-translation guided multi-round paraphrase generation framework which leverages multi- round paraphrases to improve diversity while preserving semantic information.
Outcome:	The proposed model improves diversity while preserving semantic information.

Copied to clipboard

Ruijun Chen, Jiehao Liang, Shiping Gao, Fanqi Wan, Xiaojun Quan

ICR Probe: Tracking Hidden State Dynamics for Reliable Hallucination Detection in LLMs (2025.acl-long)

Challenge:	Recent years have showcased the remarkable capabilities and performance of large language models (LLMs) across a broad range of tasks.
Approach:	They propose supervised fine-tuning (SEFT) for LLM alignment to eliminate the need for annotated samples while retaining the stability and efficiency of SFT.
Outcome:	The proposed method eliminates the need for annotated samples while maintaining the stability and efficiency of SFT.

Copied to clipboard

Zhenliang Zhang, Xinyu Hu, Huixuan Zhang, Junzhe Zhang, Xiaojun Wan

Reasoning

Challenge:	Existing methods for hallucination detection rely on static and isolated representations, overlooking their dynamic evolution across layers.
Approach:	They propose a method which captures the cross-layer evolution of hidden states and propose 'ICR Probe' which capture the evolution of the hidden states.
Outcome:	The proposed method achieves superior performance with significantly fewer parameters and ablation studies offer deeper insights into the underlying mechanism of the method, improving its interpretability.

Making Better Use of Bilingual Information for Cross-Lingual AMR Parsing (2021.findings-acl)

Copied to clipboard

Yitao Cai, Zhe Lin, Xiaojun Wan

Evaluating Self-Generated Documents for Enhancing Retrieval-Augmented Generation with Large Language Models (2025.findings-naacl)

Challenge:	Existing work on meaning representations for English and other languages finds that concepts in their predicted AMR graphs are less specific.
Approach:	They propose a cross-lingual AMR parser that can predict more precise concepts by translating translated texts and non-English texts.
Outcome:	The proposed model surpasses state-of-the-art parser by 10.6 points on Smatch F1 score.

Copied to clipboard

Jiatao Li, Xinyu Hu, Xunjian Yin, Xiaojun Wan

A Neural Approach to Pun Generation (P18-1)

Challenge:	Recent research focuses on optimizing the use of Self-Docs with their inherent properties remaining underexplored.
Approach:	They develop a taxonomy to compare the effectiveness of different types of Self-Docs and explore strategies for combining them with external sources.
Outcome:	The proposed model can supplement retrieved content and provide a powerful way to improve knowledge-intensive question answering tasks.

Copied to clipboard

Zhiwei Yu, Jiwei Tan, Xiaojun Wan

JointCQ: Improving Factual Hallucination Detection with Joint Claim and Query Generation (2026.findings-acl)

Challenge:	generating puns with artificial intelligence techniques requires manual training and templates.
Approach:	They propose neural network models for homographic pun generation that can generate puns without requiring any pun data for training.
Outcome:	The proposed models generate homographic puns of good readability and quality without training.

Copied to clipboard

Fan Xu, Huixuan Zhang, Zhenliang Zhang, Jiahao Wang, Xiaojun Wan

Continual Learning for Neural Machine Translation (2021.naacl-main)

Challenge:	Existing methods for detecting factual hallucinations in generated content exhibit limitations in the first two stages of the halluciation detection pipeline.
Approach:	They propose a joint claim-and-query generation framework that can detect factual hallucinations in generated content.
Outcome:	The proposed method outperforms existing methods on open-domain QA hallucination detection benchmarks.

Copied to clipboard

Yue Cao, Hao-Ran Wei, Boxing Chen, Xiaojun Wan

Multilinguality

Challenge:	Neural machine translation models are data-driven and require large-scale training corpus . continual learning remains a big challenge for artificial intelligence systems and models .
Approach:	They propose a continual learning framework for NMT models that incorporates multiple stages of training to alleviate catastrophic forgetting problem.
Outcome:	The proposed framework achieves superior performance compared to baseline models in all settings.

Defining and Detecting Vulnerability in Human Evaluation Guidelines: A Preliminary Study Towards Reliable NLG Evaluation (2024.naacl-long)

Copied to clipboard

Jie Ruan, WangWenqing WangWenqing, Xiaojun Wan

Guiding Abstractive Dialogue Summarization with Content Planning (2022.findings-emnlp)

Challenge:	Unreliable evaluation guidelines can yield inaccurate assessment outcomes, potentially impeding the advancement of NLG in the right direction.
Approach:	They propose to collect annotated human evaluation guidelines and a method for detecting guideline vulnerabilities using Large Language Models.
Outcome:	The proposed dataset includes eight vulnerabilities and a method for detecting guideline vulnerabilities.

Copied to clipboard

Ye Wang, Xiaojun Wan, Zhiping Cai

Exploring and Evaluating Multimodal Knowledge Reasoning Consistency of Multimodal Large Language Models (2025.findings-emnlp)

Challenge:	Existing methods for abstractive dialogue summarization struggle to maintain factual consistency between dialogue and summary.
Approach:	They propose a coarse-to-fine model for generating abstractive dialogue summaries and introduce a fact-aware reinforcement learning objective that improves the fact consistency between the dialogue and the generated summary.
Outcome:	The proposed model improves the quality of the generated summary, especially in coherence and consistency.

Copied to clipboard

Boyu Jia, Junzhe Zhang, Huixuan Zhang, Xiaojun Wan

PsyCoT: Psychological Questionnaire as Powerful Chain-of-Thought for Personality Detection (2023.findings-emnlp)

Challenge:	MLLMs have achieved significant breakthroughs in understanding across text and vision, but current models still face inconsistencies in reasoning outcomes.
Approach:	They propose to evaluate multimodal large language models using a multimodal knowledge reasoning dataset to examine the extent of consistency degradation.
Outcome:	The proposed evaluation tasks show that MLLMs are inefficient at integrating knowledge across modalities .

Copied to clipboard

Tao Yang, Tianyuan Shi, Fanqi Wan, Xiaojun Quan, Qifan Wang, Bingzhe Wu, Jiaxiang Wu

B4: A Black-Box Scrubbing Attack on LLM Watermarks (2025.naacl-long)

Challenge:	Recent advances in large language models (LLMs) have demonstrated remarkable zero-shot performance across various NLP tasks.
Approach:	They propose a method which mimics the way individuals complete psychological questionnaires in a multi-turn dialogue manner and prompts an LLM to rate individual items at each turn.
Outcome:	The proposed method improves the performance and robustness of the standard GPT-3.5 personality detection task on two benchmark datasets.

Copied to clipboard

Baizhou Huang, Xiao Pu, Xiaojun Wan

MIL-Decoding: Detoxifying Language Models at Token-Level via Multiple Instance Learning (2023.acl-long)

Challenge:	Experimental results demonstrate superior performance of black-box scrubbing attack on watermarks compared with other baselines.
Approach:	They propose a black-box scrubbing attack on watermarks that embeds a hidden pattern invisible to human into generated content of a specific LLM.
Outcome:	The proposed method outperforms baselines in 12 different environments.

Copied to clipboard

Xu Zhang, Xiaojun Wan

Error-Robust Retrieval for Chinese Spelling Check (2024.lrec-main)

Challenge:	Pre-trained language models generate toxic language which can cause security risks to their applications.
Approach:	They propose a method which detoxifies language models at token-level by interpolating it with a trained multiple instance learning network.
Outcome:	The proposed model outperforms baseline models in detoxification while hurting generation fluency a little bit.

Copied to clipboard

Xunjian Yin, Xinyu Hu, Jin Jiang, Xiaojun Wan

Information Retrieval

Challenge:	Chinese Spelling Check (CSC) aims to detect and correct spelling errors in Chinese texts . current methods may not fully leverage existing datasets, resulting in insufficient annotated data .
Approach:	They propose a plug-and-play retrieval method with error-robust information for Chinese Spelling Check . they employ multimodal representations that fuse phonetic, morphologic, and contextual information .
Outcome:	The proposed method improves on the SIGHAN benchmarks on Chinese spelling check (CSC) the proposed method is based on training data and lacks adequate parallel corpora .

Enhancing Large Language Models in Coding Through Multi-Perspective Self-Consistency (2024.acl-long)

Copied to clipboard

Baizhou Huang, Shuai Lu, Xiaojun Wan, Nan Duan

Semantic Parsing for English as a Second Language (2020.acl-main)

Challenge:	Existing work utilizes verification properties to verify and re-rank solutions in a majority voting manner, but this assumption may not hold.
Approach:	They propose a multi-perspective self-consistency framework that incorporates both inter- and intra-consistency across outputs from multiple perspectives.
Outcome:	The proposed framework significantly boosts performance of foundation models on various benchmarks, including HumanEval (+15.91%), MBPP (+6.43%) and CodeContests (+9.37%).

Copied to clipboard

Yuanyuan Zhao, Weiwei Sun, Junjie Cao, Xiaojun Wan

Diversifying Neural Text Generation with Part-of-Speech Guided Softmax and Sampling (2022.coling-1)

Challenge:	Existing studies on domain adaptation in NLP focus on learning challenges at the syntax-semantics interface during second language acquisition.
Approach:	They propose to use English Resource Grammar and TLE to parse ESL data using a reranking model to evaluate the quality of the annotations.
Outcome:	The proposed model can obtain a very promising quality in comparison to human annotations.

Copied to clipboard

Zhixian Yang, Pengxuan Xu, Xiaojun Wan

WaterPool: A Language Model Watermark Mitigating Trade-Offs among Imperceptibility, Efficacy and Robustness (2025.naacl-long)

Challenge:	Existing methods to generate text using contextual features do not consider syntactic structure clues.
Approach:	They propose using linguistic annotation, i.e., part-of-speech (POS), to guide the text generation.
Outcome:	The proposed method can generate more diverse text while maintaining comparable quality.

Copied to clipboard

Baizhou Huang, Xiaojun Wan

On the Helpfulness of Document Context to Sentence Simplification (2020.coling-main)

Challenge:	Existing methods to trace the usage of large language models often face trade-offs between imperceptibility and robustness.
Approach:	They propose a key-centered scheme to unify existing methods by decomposing a watermark into two components: a 'key module' and a "mark module".
Outcome:	The proposed method can be integrated with existing methods and achieve near-optimal imperceptibility and detection efficacy.

Copied to clipboard

Renliang Sun, Zhe Lin, Xiaojun Wan

QuantileMark: A Message-Symmetric Multi-bit Watermark for LLMs (2026.acl-long)

Challenge:	Text simplification is a hot issue in the field of natural language generation (NLG).
Approach:	They propose to use Wikipedia context to improve sentence simplification by using neural networks to learn the effects of preceding and following sentences on current sentences.
Outcome:	The proposed model outperforms the best performing model on the baseline dataset by 2.46 (7.22%).

Copied to clipboard

Junlin Zhu, Baizhou Huang, Xiaojun Wan

DivGAN: Towards Diverse Paraphrase Generation via Diversified Generative Adversarial Network (2020.findings-emnlp)

Challenge:	a number of large language models (LLMs) require multi-bit watermarking to ensure provenance.
Approach:	They propose a multi-bit watermark that embeds messages within a continuous cumulative probability interval.
Outcome:	The proposed watermark breaks message symmetry in low-entropy decoding, showing it can be used for verification and quality verification.

Copied to clipboard

Yue Cao, Xiaojun Wan

DecoCal: Decoding with Calibration in Diffusion Large Language Models (2026.acl-long)

Challenge:	Paraphrases refer to texts that convey the same meaning with different expression forms.
Approach:	They propose to incorporate a diversity loss term into a deep generative model to generate diverse paraphrases.
Outcome:	The proposed model can generate more diverse paraphrases compared with baselines.

Copied to clipboard

Fan Xu, Huixuan Zhang, Xiaojun Wan

ProactiveEval: A Unified Evaluation Framework for Proactive Dialogue Agents (2026.acl-long)

Challenge:	Diffusion Large Language Models (DLLMs) generate text via iterative token denoising . but decoding is challenging, with many tokens appearing predictable early .
Approach:	They propose a Decoding framework that performs Calibration of token-level confidence across diffusion steps and leverages the calibrated results to guide decoding decisions.
Outcome:	Experiments on multiple DLLMs and benchmarks show that DecoCal improves generation accuracy compared to existing strategies.

Copied to clipboard

Tianjian Liu, Fanqi Wan, Jiajian Guo, Xiaojun Quan

Homophonic Pun Generation with Lexically Constrained Rewriting (2020.emnlp-main)

Challenge:	Existing studies on proactive dialogue models focus on domain-specific or task-oriented scenarios, which leads to fragmented evaluations and limits the comprehensive exploration of models’ proactive dialogue abilities.
Approach:	They propose a framework for evaluating proactive dialogue capabilities of large language models that decomposes proactive dialogue into target planning and dialogue guidance, establishing evaluation metrics across various domains.
Outcome:	The proposed framework decomposes proactive dialogue into target planning and dialogue guidance, establishing evaluation metrics across various domains, and enables automatic generation of diverse and challenging evaluation data.

Copied to clipboard

Zhiwei Yu, Hongyu Zang, Xiaojun Wan

WIND: Weighting Instances Differentially for Model-Agnostic Domain Adaptation (2021.findings-acl)

Challenge:	a new method for generating puns using two homophones is needed to generate creative puns . early models for pun generation rely on templates and lack novelty.
Approach:	They propose a neural approach to generate homophonic puns with two meanings . they use constraint words to find the semantic incongruity and explicit negative constraints .
Outcome:	The proposed model achieves state-of-the-art in automatic and human evaluations.

Copied to clipboard

Xiang Chen, Yue Cao, Xiaojun Wan

Low-Resource NLP

Challenge:	Existing methods for instance weighting cannot learn the weights which make the model generalize well in target domain.
Approach:	They propose a modelagnostic instance weighting algorithm which can learn the instance weights instead of manually designed weighting metrics.
Outcome:	The proposed method can learn the instance weights instead of manually designed weighting metrics.

Explore-Instruct: Enhancing Domain-Specific Instruction Coverage through Active Exploration (2023.emnlp-main)

Copied to clipboard

Fanqi Wan, Xinting Huang, Tao Yang, Xiaojun Quan, Wei Bi, Shuming Shi

FuseChat: Knowledge Fusion of Chat Models (2025.emnlp-main)

Challenge:	Existing data for instruction-tuning are inadequate for a wide range of tasks, limiting the scope for nuanced comprehension and interactions within these domains.
Approach:	They propose to use Large Language Models to explore a multitude of variations or possibilities to improve instruction-tuning data by active exploration.
Outcome:	The proposed approach improves domain-specific instruction coverage and shows significant improvements over baselines.

Copied to clipboard

Fanqi Wan, Longguang Zhong, Ziyi Yang, Ruijun Chen, Xiaojun Quan

Who Writes What: Unveiling the Impact of Author Roles on AI-generated Text Detection (2025.acl-long)

Challenge:	Large language models (LLMs) are costly and require significant computational resources and time.
Approach:	They propose a fuse-and-merge framework for the knowledge fusion of chat LLMs . they conduct pairwise knowledge fusing on source chat LRMs to create multiple target LLM .
Outcome:	The proposed framework is superior to baselines of various sizes.

Copied to clipboard

Jiatao Li, Xiaojun Wan

Re-evaluating Automatic LLM System Ranking for Alignment with Human Preference (2025.findings-naacl)

Challenge:	Large Language Models (LLMs) require accurate text detection, but authors' characteristics are neglected.
Approach:	They investigate how author characteristics impact AI-generated text detection . they use corpus of human-authored texts and parallel AI-generated texts .
Outcome:	The results show that gender, CEFR proficiency, academic field and language environment influence detector accuracy.

Copied to clipboard

Mingqi Gao, Yixin Liu, Xinyu Hu, Xiaojun Wan, Jonathan Bragg, Arman Cohan

Challenge:	Evaluating and ranking the capabilities of different LLMs is crucial for understanding their performance and alignment with human preferences.
Approach:	They propose a system-level evaluation framework that ranks LLMs based on their alignment with human preferences.
Outcome:	The proposed framework aims to rank LLMs based on their performance and alignment with human preferences.

DialSummEval: Revisiting Summarization Evaluation for Dialogues (2022.naacl-main)

Copied to clipboard

Mingqi Gao, Xiaojun Wan

Models See Hallucinations: Evaluating the Factuality in Video Captioning (2023.emnlp-main)

Challenge:	Current models for dialogue summarization have flaws that may not be well exposed by frequently used metrics such as ROUGE.
Approach:	They propose to re-evaluate 18 categories of metrics in terms of four dimensions: coherence, consistency, fluency and relevance, as well as a unified human evaluation of various models for the first time.
Outcome:	The proposed dataset will be used to evaluate 18 categories of metrics in terms of coherence, consistency, fluency and relevance, and a unified human evaluation of various models for the first time.

Copied to clipboard

Hui Liu, Xiaojun Wan

Semantic Role Labeling for Learner Chinese: the Importance of Syntactic Parsing and L2-L1 Parallel Data (D18-1)

Challenge:	Existing evaluation metrics show little correlation with human factuality annotation.
Approach:	They propose a weakly-supervised, model-based factuality metric FactVC which outperforms previous metrics on factual evaluation of video captioning.
Outcome:	The proposed model outperforms previous metrics on factuality evaluation of video captioning.

Copied to clipboard

Zi Lin, Yuguang Duan, Yuanyuan Zhao, Weiwei Sun, Xiaojun Wan

ParaSCI: A Large Scientific Paraphrase Dataset for Longer Paraphrase Generation (2021.eacl-main)

Challenge:	a learner language (interlanguage) is an idiolect developed by a learning of a second or foreign language.
Approach:	They propose to use semantic role labeling as a case task to parse interlanguages . they then evaluate three off-the-shelf SRL systems to gauge how successful they are .
Outcome:	The proposed model achieves an F-score of 72.06, a 2.02 point improvement over the baseline.

Copied to clipboard

Qingxiu Dong, Xiaojun Wan, Yue Cao

Analyzing and Evaluating Correlation Measures in NLG Meta-Evaluation (2025.naacl-long)

Challenge:	Existing paraphrase datasets are mainly from news, novels, or social media platforms.
Approach:	They propose to build a large-scale paraphrase dataset using intra-paper and inter-paper methods . they use PDBERT as a general paraphrase discovering method to take advantage of paraphrased sentences .
Outcome:	The proposed dataset includes 33,981 paraphrase pairs from ACL and 316,063 pairs from arXiv . the major advantages of paraphrases lie in the prominent length and textual diversity .

Copied to clipboard

Mingqi Gao, Xinyu Hu, Li Lin, Xiaojun Wan

Green & Sustainable NLP

Challenge:	Existing studies have not investigated the differences between different correlation measures in meta-evaluation.
Approach:	They analyze 12 common correlation measures using real-world data from six widely-used NLG evaluation datasets and 32 evaluation metrics.
Outcome:	The proposed measures exhibit the best performance in discriminative power and ranking consistency . the measures using system-level grouping or Kendall correlation are the least sensitive to score granularity .

ReAlign: Structured Revision for Small Language Model Alignment (2025.findings-emnlp)

Copied to clipboard

Ruijun Chen, Jiajian Guo, Hongzhan Chen, Fanqi Wan, Qifan Wang, Xiaojun Quan

Automated Chess Commentator Powered by Neural Chess Engine (P19-1)

Challenge:	weak policies struggle to generate informative on-policy samples and suffer from unstable gradients when trained on off-police signals from stronger models.
Approach:	They propose a training framework that combines stability of on-policy learning with reviser-assisted supervision.
Outcome:	The proposed training framework outperforms strong preference optimization baselines on AlpacaEval-2 and Arena-Hard.

Copied to clipboard

Hongyu Zang, Zhiwei Yu, Xiaojun Wan

Point Precisely: Towards Ensuring the Precision of Data in Generated Texts Using Delayed Copy Mechanism (C18-1)

Challenge:	Existing approaches to generate chess commentary are limited in template variety and are not precise enough.
Approach:	They propose a neural chess engine into text generation models to help with encoding boards, predicting moves, and analyzing situations.
Outcome:	The proposed model can be trained to generate chess commentary texts in 5 categories . the results are both automatic and human evaluations of the model .

Copied to clipboard

Liunian Li, Xiaojun Wan

HAD: HAllucination Detection Language Models Based on a Comprehensive Hallucination Taxonomy (2026.acl-industry)

Challenge:	Recent neural generation systems have shown significant progress on data-to-text generation tasks.
Approach:	They propose a two-stage approach with a delayed copy mechanism to improve the precision of data records in the generated texts.
Outcome:	The proposed approach improves the accuracy of the generated texts on a RotoWire dataset.

Copied to clipboard

Fan Xu, Xinyu Hu, Zhenghan Yu, Li Lin, Xu Zhang, Yang Zhang, Wei Zhou, Jinjie Gu, Xiaojun Wan

TransSum: Translating Aspect and Sentiment Embeddings for Self-Supervised Opinion Summarization (2021.findings-acl)

Challenge:	relying on large language models for information has raised concerns about reliability and accuracy of outputs.
Approach:	They propose a hallucination taxonomy with 11 categories for various NLG tasks and propose HAllucination Detection models which integrate hallucinism detection, span-level identification, and correction into a single inference process.
Outcome:	The proposed models outperform baselines on HaluEval, FactCHD, and FaithBench, confirming their robustness and versatility.

Copied to clipboard

Ke Wang, Xiaojun Wan

MOVER: Mask, Over-generate and Rank for Hyperbole Generation (2022.naacl-main)

Challenge:	Existing studies focus on unsupervised opinion summarization and treat it as a normal multi-document summarizing task.
Approach:	They propose a selfsupervised opinion summarization framework TransSum that learns crucial aspect and sentiment embeddings of reviews using intra- and inter-group invariances.
Outcome:	The proposed framework outperforms baselines in generating informative, relevant and low-redundant summaries on three domains.

Copied to clipboard

Yunxiang Zhang, Xiaojun Wan

Language Generation via DAG Transduction (P18-1)

Challenge:	despite being a common figure of speech, hyperbole is under-researched in Figurative Language Processing . we use an unsupervised method to generate hyperbolic paraphrases from literal sentences .
Approach:	They propose an unsupervised method for hyperbole generation that does not require parallel literal-hyperbole pairs.
Outcome:	The proposed method outperforms baseline systems and is based on a large-scale English hyperbole corpus.

Copied to clipboard

Yajie Ye, Weiwei Sun, Xiaojun Wan

PaCoST: Paired Confidence Significance Testing for Benchmark Contamination Detection in Large Language Models (2024.findings-emnlp)

Challenge:	Existing formal frameworks for graph manipulation are underexploited.
Approach:	They propose a DAG transducer to perform graph-to-program transformation using a declarative programming language.
Outcome:	The proposed transducer achieves a BLEU-4 score of 68.07 for natural language generation from type-logical semantic graphs.

Copied to clipboard

Huixuan Zhang, Yun Lin, Xiaojun Wan

Dependency-based Mixture Language Models (2022.acl-long)

Challenge:	Large language models are trained on vast amounts of data, which may unintentionally or intentionally include data from commonly used benchmarks.
Approach:	They propose a set of requirements that practical contamination detection methods should follow to effectively detect benchmark contamination in large language models.
Outcome:	The proposed method detects whether the model is significantly more confident under the original benchmark.

Copied to clipboard

Zhixian Yang, Xiaojun Wan

Pre- and In-Parsing Models for Neural Empty Category Detection (P18-1)

Challenge:	Existing models to incorporate syntactic structures into neural language models have relied heavily on elaborate components for a specific language model, which makes them unwieldy in practice to fit into other models.
Approach:	They propose a dependency-based mixture language model that incorporates syntactic structures into neural language models by mixing previous dependency modeling probabilities with self-attention.
Outcome:	The proposed method can be easily and effectively applied to different neural language models while improving neural text generation on various tasks.

Copied to clipboard

Yufei Chen, Yuanyuan Zhao, Weiwei Sun, Xiaojun Wan

Contextual Modeling for Document-level ASR Error Correction (2024.lrec-main)

Challenge:	Existing studies on empty category detection have shown positive effects on syntactic parsing . empty categories are used to indicate long-distance dependencies, discontinuous constituents, and certain dropped elements.
Approach:	They propose to use ECD to detect empty categories without syntactic analysis.
Outcome:	The proposed models outperform the prior state-of-the-art by significant margins.

Copied to clipboard

Jin Jiang, Xunjian Yin, Xiaojun Wan, Wei Peng, Rongjun Li, Jingyuan Yang, Yanquan Zhou

Heterogeneous Graph Transformer for Graph-to-Sequence Learning (2020.acl-main)

Challenge:	Existing work on document-level ASR error correction ignores contextual information . however, there are limited studies on incorporating contextual information into AEC .
Approach:	They propose a context-aware method that retrieves contextual information from a datastore . they use two English and two Chinese datasets to model document-level AEC .
Outcome:	The proposed model can utilize contextual information to improve document-level AEC . the data store containing contextual information provides even better results .

Copied to clipboard

Shaowei Yao, Tianming Wang, Xiaojun Wan

Structured Data in NLP

Challenge:	Recent studies ignore the indirect relations between distance nodes, or treat indirect relations and direct relations in the same way.
Approach:	They propose a graph-to-sequence (Graph2Seq) encoder which models graph structure to model different relations in individual subgraphs of the original graph.
Outcome:	The proposed model outperforms the state-of-the-art on all four benchmarks of AMR-to-text generation and syntax-based neural machine translation.

Towards Document-Level Paraphrase Generation with Sentence Rewriting and Reordering (2021.findings-emnlp)

Copied to clipboard

Zhe Lin, Yitao Cai, Xiaojun Wan

Video Paragraph Captioning as a Text Summarization Task (2021.acl-short)

Challenge:	Document-level paraphrase generation is an important task in natural language processing.
Approach:	They propose a coherence relationship-guided paraphrase generation model that leverages graph GRU to encode the coherency relationship graph and get the cohesion-aware representation for each sentence.
Outcome:	The proposed model outperforms baseline models on BERTScore and diversity scores.

Copied to clipboard

Hui Liu, Xiaojun Wan

BlockPruner: Fine-grained Pruning for Large Language Models (2025.findings-acl)

Challenge:	Existing methods for video paragraph captioning use ground-truth event segments.
Approach:	They propose a video paragraph captioning task that generates coherent paragraphs without ground-truth event segments.
Outcome:	The proposed framework outperforms existing methods on two popular datasets.

Copied to clipboard

Longguang Zhong, Fanqi Wan, Ruijun Chen, Xiaojun Quan, Liangzhi Li

Accurate SHRG-Based Semantic Parsing (P18-1)

Challenge:	Large language models (LLMs) have significant computational and memory costs associated with training and inference.
Approach:	They propose a training-free structured pruning approach that targets redundancies in MHA and MLP blocks.
Outcome:	The proposed pruning approach achieves more granular and effective pruning compared to state-of-the-art pruning methods.

Copied to clipboard

Yufei Chen, Weiwei Sun, Xiaojun Wan

Retrieval-Generation Alignment for End-to-End Task-Oriented Dialogue System (2023.emnlp-main)

Challenge:	Graph-structured semantic representations can encode rich semantic information of natural language sentences.
Approach:	They propose a SHRG-based parser that relates synchronous production rules to syntacto-semantic composition processes.
Outcome:	The proposed model improves on the best existing model by 4.87 points . it relates synchronous production rules to syntacto-semantic composition process .

Copied to clipboard

Weizhou Shen, Yingqi Gao, Canbin Huang, Fanqi Wan, Xiaojun Quan, Wei Bi

Natural Language Interfaces

Challenge:	generative models struggle to distinguish subtle differences among retrieved knowledge records, resulting in suboptimal quality of generated responses.
Approach:	They propose to use maximum marginal likelihood to train a perceptive retriever by utilizing signals from response generation for supervision.
Outcome:	The proposed approach improves on three task-oriented dialogue datasets using T5 and ChatGPT as the backbone models.

Benchmarking Knowledge Boundary for Large Language Models: A Different Perspective on Model Evaluation (2024.acl-long)

Copied to clipboard

Xunjian Yin, Xu Zhang, Jie Ruan, Xiaojun Wan

Exploiting Summarization Data to Help Text Simplification (2023.eacl-main)

Challenge:	Recent advances in large language models have improved performance across tasks . however, the sensitivity of LLMs to prompt leads to unreliability of evaluation results .
Approach:	They propose a new concept to evaluate language models with a fixed question or limited paraphrases as the query.
Outcome:	The proposed method outperforms existing benchmarks on multiple language models . it avoids prompt sensitivity, rendering models more reliable and robust .

Copied to clipboard

Renliang Sun, Zhixian Yang, Xiaojun Wan

Document-Level Text Simplification: Dataset, Criteria and Baseline (2021.emnlp-main)

Challenge:	Existing text simplification datasets are limited to Wikipedia and Newsela, restricting further development of this field.
Approach:	They propose an alignment algorithm to extract sentence pairs from summarization datasets and a method to filter suitable pairs.
Outcome:	The proposed algorithm can extract sentence pairs from summarization datasets and perform well with real datasets.

Copied to clipboard

Renliang Sun, Hanqi Jin, Xiaojun Wan

Exploring Discourse Structure in Document-level Machine Translation (2023.emnlp-main)

Challenge:	Text simplification is a valuable technique, but research on it is limited.
Approach:	They propose a document-level simplification task using Wikipedia dumps as a dataset and propose an automatic evaluation metric called D-SARI.
Outcome:	The proposed metric is more suitable for document-level simplification task.

Copied to clipboard

Xinyu Hu, Xiaojun Wan

Machine Translation

Challenge:	Existing methods for document-level machine translation (DocMT) are under-utilizing the context.
Approach:	They propose a paragraph-to-paragraph translation mode that utilizes discourse information . they propose 'speech-based' translation mode which utilizes contextual information based on the context .
Outcome:	The proposed method utilizes discourse information and performs better than previous methods.

A New Benchmark and Reverse Validation Method for Passage-level Hallucination Detection (2023.findings-emnlp)

Copied to clipboard

Shiping Yang, Renliang Sun, Xiaojun Wan

Reasoning

Challenge:	Large Language Models (LLMs) are capable of working with humans in real-world scenarios, but they are prone to generate hallucinations and misinformation when deployed for mission-critical tasks.
Approach:	They propose a self-check approach to detect factual errors in a zero-resource fashion by using reverse validation to generate a hallucination detection benchmark.
Outcome:	The proposed method outperforms baseline methods while costing fewer tokens and less time.

Asking the Crowd: Question Analysis, Evaluation and Generation for Open Discussion on Online Forums (P19-1)

Copied to clipboard

Zi Chai, Xinyu Xing, Xiaojun Wan, Bo Huang

Improving Grammatical Error Correction with Data Augmentation by Editing Latent Representation (2020.coling-main)

Challenge:	Existing work on teaching machines to ask questions focused on generating fixed answers.
Approach:	They propose a model to generate open-answered questions from real-world news for open discussion . they analyze how language use affects the number of answers .
Outcome:	The proposed model generates questions with higher quality than most text generation methods.

Copied to clipboard

Zhaohong Wan, Xiaojun Wan, Wenguang Wang

Low-Resource NLP

Challenge:	Existing methods for enhancing grammatical error correction use noise to generate tokens . existing methods only generate sentences with limited error types, which leads to lack of diversity of generated errors.
Approach:	They propose a data augmentation method that can apply noise to latent representations of a sentence to generate synthetic samples with various error types.
Outcome:	The proposed method improves performance and robustness of existing models on public benchmarks and on FCE benchmarks.

Multi-Grained Knowledge Retrieval for End-to-End Task-Oriented Dialog (2023.acl-long)

Copied to clipboard

Fanqi Wan, Weizhou Shen, Ke Yang, Xiaojun Quan, Wei Bi

How to Avoid Sentences Spelling Boring? Towards a Neural Approach to Unsupervised Metaphor Generation (N19-1)

Challenge:	Existing systems blend knowledge retrieval with response generation and optimize them with direct supervision from reference responses.
Approach:	They propose a multi-grained knowledge retrieval system that decouples knowledge retrievals from response generation and introduces an entity selector and an attribute selector to acquire multigrained information from the knowledge base.
Outcome:	The proposed system performs better on small and large knowledge bases.

Copied to clipboard

Zhiwei Yu, Xiaojun Wan

DAMON: A Dialogue-Aware MCTS Framework for Jailbreaking Large Language Models (2025.emnlp-main)

Challenge:	Existing approaches to generate metaphors rely on template-based or rule-based knowledge, which constrains the diversity of generated metaphors.
Approach:	They propose a neural approach to metaphor generation that uses wiki corpus to extract metaphorically used verbs and train a language model.
Outcome:	The proposed approach generates metaphors with good readability and creativity using wiki corpus and automatic metrics and human evaluations.

Copied to clipboard

Xu Zhang, Xunjian Yin, Dinghao Jing, Huixuan Zhang, Xinyu Hu, Xiaojun Wan

Teaching the Pre-trained Model to Generate Simple Texts for Text Simplification (2023.findings-acl)

Challenge:	Existing methods for multi-turn attacks mainly utilize a predefined dialogue pattern, limiting their effectiveness in realistic situations.
Approach:	They propose a multi-turn jailbreak attack method that leverages Monte Carlo Tree Search to explore multi-turned conversational spaces and identifies sub-instruction sequences that induce harmful responses.
Outcome:	The proposed method can induce undesired behaviors across five LLMs and three datasets.

Copied to clipboard

Renliang Sun, Wei Xu, Xiaojun Wan

AMR-To-Text Generation with Graph Transformer (2020.tacl-1)

Challenge:	Existing strategies to teach pre-trained models to generate simple texts are inadequate.
Approach:	They propose a continued pre-training strategy to teach pre-trained models to generate simple texts by randomly masking text spans in ordinary texts.
Outcome:	The proposed strategy improves on lexical simplification, sentence simplification and document-level simplification tasks over existing models.

Copied to clipboard

Tianming Wang, Xiaojun Wan, Hanqi Jin

LEDOM: Reverse Language Model (2026.acl-long)

Challenge:	Abstract meaning representation (AMR)-to-text generation is challenging task for natural language processing.
Approach:	They propose a graph-to-sequence model that directly encodes AMR graphs and learns node representations.
Outcome:	The proposed model outperforms the current state-of-the-art neural approach by 1.5 BLEU points on LDC2015E86 and 4.8 BLUE points on the LDC2017T10 and achieves new state- of-the art performance.

Copied to clipboard

Xunjian Yin, Sitao Cheng, Yuxi Xie, Xinyu Hu, Li Lin, Xinyi Wang, Liangming Pan, William Yang Wang, Xiaojun Wan

A New Dataset and Empirical Study for Sentence Simplification in Chinese (2023.acl-long)

Challenge:	Autoregressive language models are trained exclusively left-to-right, yet they are limited in their ability to factorize text.
Approach:	They propose a purely reverse autoregressive language model that factorizes text as a product of left-to-right conditionals.
Outcome:	The proposed model can be used to score forward outputs using reverse posterior estimates.

Copied to clipboard

Shiping Yang, Renliang Sun, Xiaojun Wan

Automatic Generation of Citation Texts in Scholarly Papers: A Pilot Study (2020.acl-main)

Challenge:	Sentence simplification is a valuable technique that can benefit language learners and children.
Approach:	They propose a dataset for assessing sentence simplification in Chinese using manual simplifications from human annotators.
Outcome:	The proposed dataset shows that Chinese sentences are more accessible to children and nonnative readers than English sentences.

Copied to clipboard

Xinyu Xing, Xiaosheng Fan, Xiaojun Wan

Challenge:	Existing studies on automatic generation of citation texts in scholarly papers have not investigated this problem.
Approach:	They propose to train an implicit citation extraction model based on BERT and a multi-source pointer-generator network with cross attention mechanism for citation text generation.
Outcome:	The proposed model can generate short texts to describe cited papers in scholarly papers with training data.

Routing Enforced Generative Model for Recipe Generation (2020.emnlp-main)

Copied to clipboard

Zhiwei Yu, Hongyu Zang, Xiaojun Wan

Structure-Aware Pre-Training for Table-to-Text Generation (2021.findings-acl)

Challenge:	Existing methods to generate recipes with information about ingredients are difficult to use in practice.
Approach:	They propose a routing method to dive into the content selection under the internal restrictions.
Outcome:	The proposed model improves on BLEU, F1 and human evaluation.

Copied to clipboard

Xinyu Xing, Xiaojun Wan

How Do Seq2Seq Models Perform on End-to-End Data-to-Text Generation? (2022.acl-long)

Challenge:	Pretraining techniques have achieved great success on table-to-text generation.
Approach:	They propose a pre-trained model that is trained with tables and their contexts to generate fluent text from table input.
Outcome:	The proposed model can understand the structured input table and generate fluent text.

Copied to clipboard

Xunjian Yin, Xiaojun Wan

R-Bind: Unified Enhancement of Attribute and Relation Binding in Text-to-Image Diffusion Models (2025.emnlp-main)

Challenge:	Existing models for data-to-text generation are based on pipelines and end-to end architectures.
Approach:	They use multidimensional quality metrics to evaluate models on end-to-end data-totext generation and compare their performance against pipeline models.
Outcome:	The proposed model improves in Omission and Inaccuracy Extrinsic errors but increases errors such as Addition.

Copied to clipboard

Huixuan Zhang, Xiaojun Wan

Learning to Ask More: Semi-Autoregressive Sequential Question Generation under Dual-Graph Interaction (2020.acl-main)

Challenge:	Existing approaches to improve semantic binding require costly retraining or focus on only correctly generating attributes of entities. Existing methods focus on correctly generating attributes, ignoring the cruciality of correctly forming relations between entities.
Approach:	They propose a training-free method that improves both entity-attribute and entity-relation-entity binding by introducing three inference-time optimization losses that adjust attention maps during generation.
Outcome:	The proposed method improves both entity-attribute and entity-relation-entity binding without additional training.

Copied to clipboard

Zi Chai, Xiaojun Wan

Tracing Training Footprints: A Calibration Approach for Membership Inference Attacks Against Multimodal Large Language Models (2025.findings-emnlp)

Challenge:	Existing approaches to Question Generation (QG) only capture limited context dependencies due to information omission and coreference between questions.
Approach:	They propose to generate questions in a semi-autoregressive way to produce interconnected questions when there is a sequence of answers.
Outcome:	The proposed model significantly outperforms previous models on a dataset containing 81.9K questions.

Copied to clipboard

Xiaofan Zheng, Huixuan Zhang, Xiaojun Wan