Papers by Piji Li

Event Transition Planning for Open-ended Text Generation (2022.findings-acl)

Challenge:	Existing methods for social media bot detection neglect community structure and poor model generalization due to the relatively small scale of the dataset.
Approach:	They propose a framework that constructs social networks as heterogeneous graphs and uses community-aware modules to mine hard positive and hard negative samples for supervised graph contrastive learning.
Outcome:	The proposed framework outperforms baselines on three social media bot benchmarks.

Copied to clipboard

Qintong Li, Piji Li, Wei Bi, Zhaochun Ren, Yuxuan Lai, Lingpeng Kong

Understanding Social Media Cross-Modality Discourse in Linguistic Space (2022.findings-emnlp)

Challenge:	Open-ended text generation tasks require models to generate coherent continuation given limited preceding context.
Approach:	They propose a novel two-stage method which explicitly arranges ensuing events in open-ended text generation tasks.
Outcome:	The proposed method improves coherence and diversity of open-ended text generation tasks.

Copied to clipboard

Chunpu Xu, Hanzhuo Tan, Jing Li, Piji Li

Multimodality

Challenge:	Existing studies on how images are structured with texts to form coherent meanings in human cognition have not addressed the problem.
Approach:	They propose a concept of cross-modality discourse which defines how human readers couple image and text understandings.
Outcome:	The proposed model shows that trendy encoders based on multi-head attention are unable to understand cross-modality discourse and modeling texts at the output layer helps yield the-state-of-the-art results.

ReFreeKV: Towards Threshold-Free KV Cache Compression (2026.findings-acl)

Copied to clipboard

Xuanfan Ni, Liyan Xu, Chenyang Lyu, Longyue Wang, Mo Yu, Lemao Liu, Fandong Meng, Jie Zhou, Piji Li

Responsible & Trustworthy NLP

Challenge:	Towards the KV cache efficiency, we propose a new objective that lifts the threshold constraints for robust KV compression.
Approach:	They propose a method that adjusts KV cache budgets while preserving full-cache performance.
Outcome:	The proposed method can reduce memory consumption while preserving full-cache performance.

Improve Language Model and Brain Alignment via Associative Memory (2025.findings-acl)

Copied to clipboard

Congchi Yin, Yongpeng Zhang, Xuyun Wen, Piji Li

CLINE: Contrastive Learning with Semantic Negative Examples for Natural Language Understanding (2021.acl-long)

Challenge:	Existing studies have shown that associative memory is essential for language comprehension and comprehension.
Approach:	They propose to integrate associative memory into language models to improve alignment . they find alignment is improved in brain regions closely related to associativ memory processing .
Outcome:	The proposed model improves in brain regions closely related to associative memory processing.

Copied to clipboard

Dong Wang, Ning Ding, Piji Li, Haitao Zheng

HELPD: Mitigating Hallucination of LVLMs by Hierarchical Feedback Learning with Vision-enhanced Penalty Decoding (2024.emnlp-main)

Challenge:	Pre-trained language models are vulnerable to simple perturbations, causing poor robustness . recent studies show that adversarial training is useless or harmful for the model to detect these semantic changes.
Approach:	They propose to use adversarial training to improve the robustness of pre-trained models . they propose to construct negative examples with similar and opposite semantics .
Outcome:	Empirical results show that the proposed approach improves on sentiment analysis, reasoning, and reading comprehension tasks.

Copied to clipboard

Fan Yuan, Chi Qin, Xiaogang Xu, Piji Li

Multimodality

Challenge:	Existing work detects hallucination by directly judging whether an object exists in an image, overlooking the association between the object and semantics.
Approach:	They propose a framework that incorporates hallucination feedback at both object and sentence semantic levels to alleviate over 15% of hallucinism.
Outcome:	The proposed framework can alleviate over 15% of hallucination even with a marginal degree of training.

Generating Diverse Training Samples for Relation Extraction with Large Language Models (2025.acl-long)

Copied to clipboard

Zexuan Li, Hongliang Dai, Piji Li

Exclusive Hierarchical Decoding for Deep Keyphrase Generation (2020.acl-main)

Challenge:	Existing models for Relation Extraction (RE) have good results on many benchmarks, but data scarcity is a common problem.
Approach:	They propose to use Large Language Models to generate training data for Relation Extraction . they propose to make LLMs produce dissimilar samples by direct instruction .
Outcome:	The proposed approach improves the diversity of training samples generated with LLMs while maintaining correctness.

Copied to clipboard

Wang Chen, Hou Pong Chan, Piji Li, Irwin King

Structured Data in NLP

Challenge:	Existing approaches to generate keyphrases ignore hierarchical compositionality of keyphrase set and generate duplicated keyphrase sets.
Approach:	They propose a hierarchical decoding framework that explicitly models hierarchic compositionality of a keyphrase set and either a soft or a hard exclusion mechanism to enhance the diversity of the generated keyphrases.
Outcome:	The proposed framework generates less duplicated and more accurate keyphrases on a set of keyphrase sets.

Multi-Source Multi-Type Knowledge Exploration and Exploitation for Dialogue Generation (2023.emnlp-main)

Copied to clipboard

Xuanfan Ni, Hongliang Dai, Zhaochun Ren, Piji Li

Dialogue Systems & Conversational Agents

Challenge:	Existing models focus on identifying specific types of dialogue knowledge and utilizing corresponding datasets for training, but lack generalization capabilities and computational resources.
Approach:	They propose a framework that explores multi-source multi-type knowledge from LLMs by leveraging diverse datasets and exploits it for response generation.
Outcome:	The proposed framework exploits multi-source multi-type knowledge from LLMs to generate coherent, informative, and fluent responses.

Tail-to-Tail Non-Autoregressive Sequence Prediction for Chinese Grammatical Error Correction (2021.acl-long)

Copied to clipboard

Piji Li, Shuming Shi

Syntactic Text Processing

Challenge:	Experimental results demonstrate the effectiveness of Tail-to-Tail (TtT) non-autoregressive sequence prediction for Chinese Grammatical Error Correction (CGEC)
Approach:	They propose a framework for Chinese Grammatical Error Correction (CGEC) that uses a BERT-initialized Transformer Encoder to model the error positions.
Outcome:	The proposed framework solves the problem of Chinese Grammatical Error Correction (CGEC) by modeling the token dependencies.

Leveraging Label Semantics and Entity Description Generation for LLM-based Fine-grained Entity Typing (2026.findings-acl)

Copied to clipboard

Yingying Ma, Hongliang Dai, Xia Zhang, Piji Li

Information Extraction & Text Mining

Challenge:	Fine-grained entity typing (FET) aims to assign semantically rich and contextually appropriate types to entity mentions.
Approach:	They propose a descriptor-based retrieval-augmented framework that reduces effective label space . they propose to use natural language descriptores as an intermediate semantic representation .
Outcome:	The proposed framework outperforms existing methods under noisy supervision.

uChecker: Masked Pretrained Language Models as Unsupervised Chinese Spelling Checkers (2022.coling-1)

Copied to clipboard

Piji Li

Topic-Guided Self-Introduction Generation for Social Media Users (2023.findings-acl)

Challenge:	Chinese Spelling Check (CSC) is a crucial and essential task in the area of natural language processing.
Approach:	They propose a framework to conduct unsupervised spelling error detection and correction using masked pretrained language models such as BERT.
Outcome:	The proposed model improves character-level and sentence-level accuracy, precision, recall, and F1-Measure on standard datasets.

Copied to clipboard

Chunpu Xu, Jing Li, Piji Li, Min Yang

Dialogue Generation on Infrequent Sentence Functions via Structured Meta-Learning (2020.findings-emnlp)

Challenge:	Existing studies on social media use tags to profile users, but we have found that sentence-level self-introductions are more natural and engaging.
Approach:	They propose a novel topic-guided encoder-decoder framework that uses a user's tweeting history to generate a short sentence outlining their personal interests.
Outcome:	The proposed framework outperforms existing encoder-decoder models on a large-scale Twitter dataset and shows that it is more natural and engaging than previous approaches.

Copied to clipboard

Yifan Gao, Piji Li, Wei Bi, Xiaojiang Liu, Michael Lyu, Irwin King

Dialogue Systems & Conversational Agents

Challenge:	Sentence function is an important linguistic feature indicating the communicative purpose of a sentence in a conversation.
Approach:	They propose a structured meta-learning approach for dialogue generation on infrequent sentence functions.
Outcome:	The proposed approach improves informativeness and relevance of dialogue generation on infrequent sentence functions while preserving knowledge generalization for similar sentence functions.

Contrastive Representation Learning for Exemplar-Guided Paraphrase Generation (2021.findings-emnlp)

Copied to clipboard

Haoran Yang, Wai Lam, Piji Li

Representation Learning

Challenge:	Exemplar-Guided Paraphrase Generation (EGPG) aims to generate a target sentence which conforms to the style of the given exemplar while encapsulating the content information of the source sentence.
Approach:	They propose a method to generate a target sentence which conforms to the style of the given exemplar while encapsulating the content information of the source sentence.
Outcome:	The proposed method can generate a target sentence which conforms to the style of the given exemplar while encapsulating the content information of the source sentence.

Semantic are Beacons: A Semantic Perspective for Unveiling Parameter-Efficient Fine-Tuning in Knowledge Learning (2024.findings-acl)

Copied to clipboard

Renzhi Wang, Piji Li

Tackling Long-Tailed Relations and Uncommon Entities in Knowledge Graph Completion (D19-1)

Challenge:	Parameter-Efficient Fine-Tuning (PEFT) methods allow efficient adaptation of Large Language Models (LLMs) to various downstream tasks, but their effectiveness diminishes when downstream tasks require accurate learning of specific knowledge.
Approach:	They propose a method that fine-tunes a limited number of model parameters while keeping the majority of original parameters fixed.
Outcome:	The proposed method is able to perform on open-source large language models and validates the semantic challenge in PEFT.

Copied to clipboard

Zihao Wang, Kwunping Lai, Piji Li, Lidong Bing, Wai Lam

CRISP: Compressing Redundancy in Chain-of-Thought via Intrinsic Saliency Pruning (2026.findings-acl)

Challenge:	Recent studies have focused on the large proportion of infrequent relations which have been ignored by previous studies.
Approach:	They propose a meta-learning framework that aims at handling infrequent relations with few-shot learning and uncommon entities by using textual descriptions.
Outcome:	The proposed framework outperforms existing methods when dealing with infrequent relations and uncommon entities.

Copied to clipboard

Yangsong Lan, Hongliang Dai, Piji Li

Responsible & Trustworthy NLP

Challenge:	Existing approaches to compress CoT with external compressors fail to align with the model’s internal reasoning dynamics, resulting in the loss of critical logical steps.
Approach:	They propose a framework that exploits the model’s intrinsic saliency to compress CoT by exploiting its reasoning termination token .
Outcome:	The proposed framework reduces redundancy in reasoning chain by exploiting the model’s intrinsic saliency.

An Integrated Approach for Keyphrase Generation via Exploring the Power of Retrieval and Extraction (N19-1)

Copied to clipboard

Wang Chen, Hou Pong Chan, Piji Li, Lidong Bing, Irwin King

Information Retrieval

Challenge:	Existing methods on keyphrase generation are purely extractive or generative . however, extractive methods cannot predict absent keyphrases which are not in the document.
Approach:	They propose a multi-task learning framework that jointly learns an extractive model and a generative model.
Outcome:	The proposed approach outperforms the state-of-the-art methods on five keyphrase generation tasks.

LEMoE: Advanced Mixture of Experts Adaptor for Lifelong Model Editing of Large Language Models (2024.emnlp-main)

Copied to clipboard

Renzhi Wang, Piji Li

Language Reconstruction with Brain Predictive Coding from fMRI Data (2026.acl-long)

Challenge:	Existing methods for single and batch model editing fail to apply or perform sub-optimally when faced with lifelong model editing.
Approach:	They propose an advanced Mixture of Experts (MoE) adaptor for lifelong model editing that incorporates a novel KV anchor routing method to enhance routing consistency between training and inference stage.
Outcome:	The proposed method surpasses existing models while maintaining outstanding performance in batch editing task.

Copied to clipboard

Congchi Yin, Ziyi Ye, Piji Li

Linguistic Theories

Challenge:	Existing studies have shown that the perception of speech can be decoded from brain signals and subsequently reconstructed as continuous language.
Approach:	They propose to use FMRI-to-text decoding with Predictive coding to generate a main network and a side network to generate brain predictive representations from related regions of interest.
Outcome:	The proposed model outperforms current decoding models on several evaluation metrics on two naturalistic language comprehension fMRI datasets.

Interconnected Question Generation with Coreference Alignment and Conversation Flow Modeling (P19-1)

Copied to clipboard

Yifan Gao, Piji Li, Irwin King, Michael R. Lyu

Natural Language Interfaces

Challenge:	Extensive experiments show that our system outperforms several baselines and can generate highly conversational questions.
Approach:	They propose a neural model that generates interconnected questions in question-answering style conversations.
Outcome:	The proposed model outperforms baselines and can generate highly conversational questions.

InfoDiffusion: Information Entropy Aware Diffusion Process for Non-Autoregressive Text Generation (2023.findings-emnlp)

Copied to clipboard

Renzhi Wang, Jing Li, Piji Li

Rethinking Cross-Subject Data Splitting for Brain-to-Text Decoding (2025.emnlp-main)

Challenge:	Existing text diffusion models have failed to capture the difference between the “easy-first” process and the “keyword-first" process of humans.
Approach:	They propose a non-autoregressive text diffusion model that incorporates a "keyinfo-first" generation strategy and a noise schedule based on the amount of text information.
Outcome:	The proposed model outperforms the baseline model in terms of generation quality and diversity, and higher sampling efficiency.

Copied to clipboard

Congchi Yin, Qian Yu, Zhiwei Fang, Changping Peng, Piji Li

A Training-free and Reference-free Summarization Evaluation Metric via Centrality-weighted Relevance and Self-referenced Redundancy (2021.acl-long)

Challenge:	Recent studies have successfully decoded natural language from non-invasive brain signals . current dataset splitting methods suffer from data leakage problem .
Approach:	They propose a right cross-subject data splitting criterion without data leakage for decoding fMRI and EEG signal to text.
Outcome:	The proposed method overfits and overestimates brain-to-text decoding models.

Copied to clipboard

Wang Chen, Piji Li, Irwin King

Decoding the Echoes of Vision from fMRI: Memory Disentangling for Past Semantic Information (2024.emnlp-main)

Challenge:	Existing evaluation metrics for text summarization systems are expensive and time-consuming.
Approach:	They propose a training-free and reference-free summarization evaluation metric that incorporates a centrality-weighted relevance score and a self-referenced redundancy score.
Outcome:	The proposed evaluation metric outperforms existing methods on multi-document and single-document summarization evaluation.

Copied to clipboard

Runze Xia, Congchi Yin, Piji Li

Visual Data in NLP

Challenge:	Experimental results demonstrate that this method effectively disentangles the information within fMRI signals.
Approach:	They propose a task Memory Disentangling which extracts and decodes past information from fMRI signals.
Outcome:	The proposed method extracts and decodes past information from fMRI signals.

Effidit: An Assistant for Improving Writing Efficiency (2023.acl-demo)

Copied to clipboard

Shuming Shi, Enbo Zhao, Wei Bi, Deng Cai, Leyang Cui, Xinting Huang, Haiyun Jiang, Duyu Tang, Kaiqiang Song, Longyue Wang, Chenyan Huang, Guoping Huang, Yan Wang, Piji Li

Green & Sustainable NLP

Challenge:	Effidit is a digital writing assistant that provides three modules to help users write faster and more efficiently.
Approach:	They present Effidit, a digital writing assistant that provides three modules to help users write higher-quality text more efficiently.
Outcome:	Effidit expands the capabilities of a typical writing assistant by providing three modules . Effit can help users create their own text faster and more efficiently .

Keep it Consistent: Topic-Aware Storytelling from an Image Stream via Iterative Multi-agent Communication (2020.coling-main)

Copied to clipboard

Ruize Wang, Zhongyu Wei, Ying Cheng, Piji Li, Haijun Shan, Ji Zhang, Qi Zhang, Xuanjing Huang

Multimodality

Challenge:	Existing methods for visual storytelling construct text description independently for each image and roughly concatenate them as a story, which leads to the problem of generating semantically incoherent content.
Approach:	They propose a topic description task to detect the global semantic context of an image stream and a story is then constructed with the guidance of the topic description.
Outcome:	The proposed framework can generate stories with higher quality compared to state-of-the-art methods on a VIST dataset.

How to Write Summaries with Patterns? Learning towards Abstractive Summarization through Prototype Editing (D19-1)

Copied to clipboard

Shen Gao, Xiuying Chen, Piji Li, Zhangming Chan, Dongyan Zhao, Rui Yan

Rigid Formats Controlled Text Generation (2020.acl-main)

Challenge:	Extensive experiments on a large-scale real-world text summarization dataset show that PESG achieves the state-of-the-art performance in terms of both automatic metrics and human evaluations.
Approach:	They propose a model that learns summary patterns and prototype facts from a prototype document . they use a fact checker to estimate mutual information between the input document and generated summary .
Outcome:	Experiments on a large-scale real-world text summarization dataset show that PESG achieves state-of-the-art performance.

Copied to clipboard

Piji Li, Haisong Zhang, Xiaojiang Liu, Shuming Shi

Semi-supervised Text Style Transfer: Cross Projection in Latent Space (D19-1)

Challenge:	Neural text generation is a challenging task that requires rigid formats to be controlled . a framework called SongNet is designed to tackle this problem .
Approach:	They propose a framework to tackle a task called rigid formats controlled text generation . they propose rhyming schemes and a transformer-based auto-regressive language model to improve the modeling performance .
Outcome:	The proposed framework improves the performance on format, rhyme, and sentence integrity.

Copied to clipboard

Mingyue Shang, Piji Li, Zhenxin Fu, Lidong Bing, Dongyan Zhao, Shuming Shi, Rui Yan

“Is Whole Word Masking Always Better for Chinese BERT?”: Probing on Chinese Grammatical Error Correction (2022.findings-acl)

Challenge:	Text style transfer task has long suffered from the shortage of parallel data .
Approach:	They propose a semi-supervised text style transfer model that combines parallel data with large-scale nonparallel data to train it.
Outcome:	The proposed model can transfer a sentence of one style to another while retaining its original content meaning while preserving its original meaning.

Copied to clipboard

Yong Dai, Linyang Li, Cong Zhou, Zhangyin Feng, Enbo Zhao, Xipeng Qiu, Piji Li, Duyu Tang

Bridging the Gap between Pre-Training and Fine-Tuning for Commonsense Generation (2023.findings-eacl)

Challenge:	a Chinese model with whole word masking has no subword because each token is an atomic character.
Approach:	They propose to use whole word masking to mask all subwords corresponding to a word at once . they ask models to revise or insert tokens in a masked language modeling manner .
Outcome:	The proposed model performs better when one character is inserted or replaced . the model trained with standard character-level masking performs best when one token is masked .

Copied to clipboard

Haoran Yang, Yan Wang, Piji Li, Wei Bi, Wai Lam, Chen Xu

DUD: Decoupled Update Dynamics for Reliable Uncertainty Quantification in Large Language Models (2026.acl-long)

Challenge:	Existing methods focusing on this task usually concatenate the concatened concepts words as the inputs of a pre-trained language model (PLM) however, in pre-training, the input is often corrupted sentences with correct word order.
Approach:	They propose a two-stage framework to improve the ability of pre-trained language models to deal with masked sentences with incorrect word order and a special token to make the input distribution more similar to the one used in pre-training.
Outcome:	The proposed method is able to generate a sentence containing all given concepts and correctly describe the relations between concepts.

Copied to clipboard

Yixin Bu, Runze Xia, Guanyun Zou, Yupeng Ji, Hongliang Dai, Haodong Liu, Piji Li

M-BRe: Discovering Training Samples for Relation Extraction from Unlabeled Texts with Large Language Models (2025.emnlp-main)

Challenge:	Accurate Uncertainty Quantification (UQ) is critical for reliable deployment of Large Language Models (LLMs).
Approach:	They propose a framework that explicitly decouples FFN and Attention contributions via noise-induced causal interventions to capture model's internal fragility.
Outcome:	The proposed framework outperforms state-of-the-art baselines in both uncertainty estimation and calibration while exhibiting superior cross-dataset generalization.

Copied to clipboard

Zexuan Li, Hongliang Dai, Piji Li

Information Extraction & Text Mining

Challenge:	Existing methods to extract training instances from unlabeled texts are expensive . sentences that contain the target relations in texts can be scarce and difficult to find .
Approach:	They propose a framework that can automatically extract training instances from unlabeled texts for RE.
Outcome:	The proposed method can extract training instances from unlabeled texts for RE.

Slot Dependency Modeling for Zero-Shot Cross-Domain Dialogue State Tracking (2022.coling-1)

Copied to clipboard

Qingyue Wang, Yanan Cao, Piji Li, Yanhe Fu, Zheng Lin, Li Guo

Low-Resource NLP

Challenge:	Existing zero-shot learning methods ignore slot dependencies in a multidomain dialogue . experimental results show the effectiveness of our proposed method over existing state-of-art generation methods .
Approach:	They propose to use slot prompts combination, slot values demonstration and slot constraint object to model slot-slot dependency, slot-value dependency and slot-context dependency respectively.
Outcome:	The proposed method outperforms state-of-the-art methods under zero-shot/few-shot settings.

Characteristic AI Agents via Large Language Models (2024.lrec-main)

Copied to clipboard

Xi Wang, Hongliang Dai, Shen Gao, Piji Li