Papers by Richong Zhang

E-VarM: Enhanced Variational Word Masks to Improve the Interpretability of Text Classification Models (2022.coling-1)

Challenge:	Existing semi-supervised text classification methods suffer from categorical boundary issues . existing methods suffer by ambiguous categoric boundaries, making it difficult to generate reliable pseudo-labels for each category.
Approach:	They propose a semi-supervised framework that assigns pseudo-labels to unlabeled data . they exploit categorical prototypes to assimilate instance representations within the same category .
Outcome:	Empirical studies show that the proposed framework is effective . it uses prototypical cluster separation and prototypical-center data selection .

Copied to clipboard

Ling Ge, ChunMing Hu, Guanghui Ma, Junshuang Wu, Junfan Chen, JiHong Liu, Hong Zhang, Wenyi Qin, Richong Zhang

Explainability & Interpretability in NLP

Challenge:	Empirical studies show that our approach outperforms the SOTA methods in improving the interpretability of text classification models.
Approach:	They propose an enhanced variational word masks approach that exploits the Variational Information Bottleneck to obtain task-specific words.
Outcome:	Empirical results show that the proposed method outperforms the SOTA methods in improving the interpretability of the model.

Neural Dialogue State Tracking with Temporally Expressive Networks (2020.findings-emnlp)

Copied to clipboard

Junfan Chen, Richong Zhang, Yongyi Mao, Jie Xu

Dialogue Systems & Conversational Agents

Challenge:	Existing models ignore temporal feature dependencies across dialogue turns or fail to explicitly model temporal state dependencies in a dialogue.
Approach:	They propose to combine temporal feature dependencies in spoken dialogues by using recurrent networks and probabilistic graphical models.
Outcome:	The proposed model improves turn-level-state prediction and state aggregation on standard datasets.

Improving Data Annotation for Low-Resource Relation Extraction with Logical Rule-Augmented Collaborative Language Models (2025.naacl-long)

Copied to clipboard

Xiyang Liu, Chunming Hu, Richong Zhang, Junfan Chen, Baowen Xu

Language Models

Challenge:	Low-resource relation extraction aims to identify semantic relationships using scarce labeled data.
Approach:	They propose a framework that iteratively integrates high-confidence predictions of rule-enhanced relation extractors with varying scales to obtain reliable pseudo annotations from massive unlabeled samples without human supervision.
Outcome:	The proposed framework achieves state-of-the-art on benchmark datasets in few-shot scenarios.

DropMix: A Textual Data Augmentation Combining Dropout with Mixup (2022.emnlp-main)

Copied to clipboard

Fanshuang Kong, Richong Zhang, Xiaohui Guo, Samuel Mensah, Yongyi Mao

Low-Resource NLP

Challenge:	Existing methods to overcome overfitting in text learning do not consider dimensionality . dimensionalization is important for deep neural networks to overcome the problem .
Approach:	They propose a saliency map-based approach to overcome overfitting in text learning . they propose augmentation regularization methods such as Dropout and Mixup to improve regularization .
Outcome:	Empirical results show that the proposed approach overcomes overfitting in text learning . dropout and mixup methods are effective in enhancing regularization .

Learning VAE-LDA Models with Rounded Reparameterization Trick (2020.emnlp-main)

Copied to clipboard

Runzhi Tian, Yongyi Mao, Richong Zhang

Text Style Transferring via Adversarial Masking and Styled Filling (2022.emnlp-main)

Challenge:	No reparameterization form of Dirichlet distributions is known to date for topic models .
Approach:	They propose a method to reparameterize Dirichlet distributions for the learning of VAE-LDA models by using a latent Dirichlets prior.
Outcome:	The proposed method outperforms existing neural topic models on benchmark datasets and on a synthetic dataset.

Copied to clipboard

Jiarui Wang, Richong Zhang, Junfan Chen, Jaein Kim, Yongyi Mao

Robustness in NLP

Challenge:	Existing models for text style transfer suffer from two challenges: the word masking procedure may mistakenly remove unexpected words and the selected words in the word filling procedure lack diversity and semantic consistency.
Approach:	They propose a style transfer model with adversarial masking and styled filling techniques to solve these challenges.
Outcome:	The proposed model performs well on two benchmark text style transfer data sets.

Contrastive Learning with Expectation-Maximization for Weakly Supervised Phrase Grounding (2022.emnlp-main)

Copied to clipboard

Keqin Chen, Richong Zhang, Samuel Mensah, Yongyi Mao

Uncover the Ground-Truth Relations in Distant Supervision: A Neural Expectation-Maximization Framework (D19-1)

Challenge:	Weakly supervised phrase grounding aims to learn an alignment between phrases in a caption and objects in an image using only caption-image annotations.
Approach:	They propose a novel contrastive learning framework that adaptively refines the target prediction by using only caption-image annotations.
Outcome:	The proposed framework outperforms existing methods on two widely used benchmarks, Flickr30K Entities and RefCOCO+.

Copied to clipboard

Junfan Chen, Richong Zhang, Yongyi Mao, Hongyu Guo, Jie Xu

Challenge:	Existing methods for relation extraction assume that text is noisy, but its corresponding labels are clean.
Approach:	They propose a framework that combines neural network and probabilistic modelling to denoise noisy relation labels.
Outcome:	The proposed framework improves the current art in uncovering the ground-truth relation labels.

Explicit Role Interaction Network for Event Argument Extraction (2022.findings-emnlp)

Copied to clipboard

Nan Ding, Chunming Hu, Kai Sun, Samuel Mensah, Richong Zhang

LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Models (2024.acl-demos)

Challenge:	Existing methods extract arguments of each role independently, ignoring the relationship between different roles.
Approach:	They propose a neural model that captures the correlations between different argument roles within an event.
Outcome:	Extensive experiments on the benchmark dataset ACE2005 show the superiority of the proposed model over existing methods.

Copied to clipboard

Yaowei Zheng, Richong Zhang, Junhao Zhang, YeYanhan YeYanhan, Zheyan Luo

An Unsupervised Multiple-Task and Multiple-Teacher Model for Cross-lingual Named Entity Recognition (2022.acl-long)

Challenge:	Efficient fine-tuning of large language models requires non-trivial efforts to implement these methods on different models.
Approach:	They propose a framework that democratizes the fine-tuning of large language models by integrating a suite of efficient training methods into one framework.
Outcome:	The proposed framework is able to scale to 100+ LLMs without coding and receives over 25,000 stars and 3,000 forks.

Copied to clipboard

Zhuoran Li, Chunming Hu, Xiaohui Guo, Junfan Chen, Wenyi Qin, Richong Zhang

A Transformational Biencoder with In-Domain Negative Sampling for Zero-Shot Entity Linking (2022.findings-acl)

Challenge:	Existing models for named entity recognition only consider the potential transferability between two identical tasks across both domains.
Approach:	They propose to use a similarity metric model to improve cross-lingual named entity recognition task on target domain.
Outcome:	Empirical studies on 7 different languages confirm the effectiveness of the proposed model.

Copied to clipboard

Kai Sun, Richong Zhang, Samuel Mensah, Yongyi Mao, Xudong Liu

Tucker Decomposition with Frequency Attention for Temporal Knowledge Graph Completion (2023.findings-acl)

Challenge:	Recent work on entity linking has focused on the zero-shot scenario where at test time the entity mention to be labelled is never seen during training.
Approach:	They propose a transformational biencoder that integrates a transform into BERT to perform a zero-shot transfer from the source domain to the target domain.
Outcome:	The proposed model performs a zero-shot transfer from the source domain to the target domain on a benchmark dataset and achieves new state-of-the-art.

Copied to clipboard

Likang Xiao, Richong Zhang, Zijie Chen, Junfan Chen

Anaphor Assisted Document-Level Relation Extraction (2023.emnlp-main)

Challenge:	Existing models for temporal knowledge graph completion only consider the combination of one relation with one timestamp, ignoring the global nature of the embedding.
Approach:	They propose a temporal knowledge Graph Completion model that captures global temporal dependencies between one relation and the entire timestamp.
Outcome:	The proposed model outperforms the state-of-the-art models on three standard TKGC datasets on several metrics.

Copied to clipboard

Chonggang Lu, Richong Zhang, Kai Sun, Jaein Kim, Cunwang Zhang, Yongyi Mao

Aspect-Level Sentiment Analysis Via Convolution over Dependency Tree (D19-1)

Challenge:	Existing methods for document-level relation extraction are incomplete and lack anaphor for identifying relations between entities.
Approach:	They propose an Anaphor-Assisted (AA) framework for document-level relation extraction . they use a document or sentences as intermediate nodes to model cross-sentence entity interactions .
Outcome:	The proposed framework achieves state-of-the-art on the widely-used datasets.

Copied to clipboard

Kai Sun, Richong Zhang, Samuel Mensah, Yongyi Mao, Xudong Liu

Sentiment Analysis

Challenge:	Existing methods to identify sentiment polarity of opinion words are cumbersome due to the amount of opinionated material on the internet.
Approach:	They propose a method to identify sentiment polarity of opinion words on a specific aspect of a sentence using neural networks.
Outcome:	The proposed method is the state-of-the-art in aspect-based sentiment classification.

Easy Dataset: A Unified and Extensible Framework for Synthesizing LLM Fine-Tuning Data from Unstructured Documents (2025.emnlp-demos)

Copied to clipboard

Ziyang Miao, Qiyu Sun, Jingyuan Wang, Yuchen Gong, Yaowei Zheng, Shiqi Li, Richong Zhang

Language Models

Challenge:	Existing data synthesis tools struggle to extract reliable fine-tuning data from heterogeneous documents.
Approach:	They propose a framework for synthesizing fine-tuning data from unstructured documents via an intuitive graphical user interface.
Outcome:	The proposed framework can extract reliable data from unstructured documents via an intuitive graphical user interface (GUI) it leverages persona-driven prompting approach to generate diverse question-answer pairs using public-available LLMs.

A Text is Worth Several Tokens: Text Embedding from LLMs Secretly Aligns Well with The Key Tokens (2025.acl-long)

Copied to clipboard

Zhijie Nie, Richong Zhang, Zhanyu Wu

Adversarial Metric Learning for Fine-Grained Emotion Classification (2026.acl-long)

Challenge:	et al., 2023) show that text embeddings from large language models can be aligned with key tokens in input text.
Approach:	They propose a sparse retrieval method based on aligned tokens for large language models . they show that this phenomenon is universal and is not affected by model architecture .
Outcome:	The proposed method can achieve 80% of the dense retrieval effect of the same model while reducing the computation significantly.

Copied to clipboard

Junfan Chen, Sizhe Wu, Richong Zhang, Chunming Hu

Robustness in NLP

Challenge:	Recent advances in fine-grained emotion classification relied on contrastive learning with hard-pair mining.
Approach:	They propose an adversarial metric learning framework that replaces fixed similarity metrics with a learnable metric family and trains representations to remain discriminative under worst-case similarity distortions.
Outcome:	The proposed framework trains a pairwise discriminator to maximally confuse two hard pair types while training the encoder to remain discriminative under worst-case similarity distortions.

Hypernym Discovery via a Recurrent Mapping Model (2021.findings-acl)

Copied to clipboard

Yuhang Bai, Richong Zhang, Fanshuang Kong, Junfan Chen, Yongyi Mao

Challenge:	Empirical studies on SemEval-2018 Task 9 confirm the effectiveness of the presented model.
Approach:	They propose a parallel style model that maps query words to their hypernyms . they use a lexical-semantic relation to name a specific instance or subtype hyponym .
Outcome:	Empirical results on SemEval-2018 Task 9 confirm the effectiveness of the proposed model.

Open-Set Semi-Supervised Text Classification via Adversarial Disagreement Maximization (2024.acl-long)

Copied to clipboard

Junfan Chen, Richong Zhang, Junchi Chen, Chunming Hu

Progressively Modality Freezing for Multi-Modal Entity Alignment (2024.acl-long)

Challenge:	Open-Set Semi-Supervised Text Classification (OSTC) aims to train a classification model on a limited set of labeled texts along with plenty of unlabeled examples.
Approach:	They propose to train a classification model on a limited set of labeled texts alongside plenty of unlabeled examples that include both in-distribution and out-of-difference examples.
Outcome:	The proposed model improves on outlier detection and abnormal example detection and calibration.

Copied to clipboard

Yani Huang, Xuefeng Zhang, Richong Zhang, Junfan Chen, Jaein Kim

Multimodality

Challenge:	Multi-Modal Entity Alignment aims to discover identical entities across heterogeneous knowledge graphs.
Approach:	They propose a strategy of progressive modality freezing that focuses on alignment-relevant features and enhances multi-modal feature fusion.
Outcome:	The proposed approach demonstrates state-of-the-art performance and the rationale for freezing modalities.

Syntax Encoding with Application in Authorship Attribution (D18-1)

Copied to clipboard

Richong Zhang, Zhiyuan Hu, Hongyu Guo, Yongyi Mao

Representation Learning

Challenge:	Existing approaches to extract syntactic features from text or sentences are limited by the loss of rich structural information contained in the syntax tree.
Approach:	They propose to embed the syntax parse tree of sentence into a learnable distributed representation . they show that the approach improves upon the prior art and achieves new performance records .
Outcome:	The proposed approach improves upon the prior art and achieves new performance records on five benchmarking data sets.

Recurrent Interaction Network for Jointly Extracting Entities and Classifying Relations (2020.emnlp-main)

Copied to clipboard

Kai Sun, Richong Zhang, Samuel Mensah, Yongyi Mao, Xudong Liu

Calibrating Pseudo-Labeling with Class Distribution for Semi-supervised Text Classification (2025.emnlp-main)

Challenge:	Existing methods to solve the extraction problem learn interactions between the two tasks through a shared network .
Approach:	They propose to use multi-task learning to address the joint extraction of entity and relation . they exploit correlation between ER and relation classification tasks to improve performance .
Outcome:	Empirical results show that the proposed model improves on two real-world datasets.

Copied to clipboard

Weiyi Yang, Richong Zhang, Junfan Chen, Jiawei Sheng

Parameter-free Automatically Prompting: A Latent Pseudo Label Mapping Model for Prompt-based Learning (2022.findings-emnlp)

Challenge:	Existing studies develop effective pseudo-labeling methods, but they struggle with unlabeled data that have imbalanced classes mismatched with the labeled data.
Approach:	They propose to use pseudo-labeling to train text classification models with few labeled data and massive unlabeled data.
Outcome:	Empirical results show that the proposed model outperforms state-of-the-art methods on 3 common benchmarks.

Copied to clipboard

Jirui Qi, Richong Zhang, Junfan Chen, Jaein Kim, Yongyi Mao

Learning to Select: Query-Aware Adaptive Dimension Selection for Dense Retrieval (2026.acl-long)

Challenge:	Existing manual label mapping methods that require extra parameters and human knowledge are limited in data.
Approach:	They propose a Latent Pseudo Label Mapping method that optimizes the label mapping without human knowledge and extra parameters.
Outcome:	The proposed method outperforms the standard SOTA method in few-shot learning tasks and significantly outperformed the standard ALM method which requires extra task-specific prior knowledge.

Copied to clipboard

Zhanyu Wu, Richong Zhang, Zhijie Nie

Dynamic Task Vector Grouping for Efficient Multi-Task Prompt Tuning (2025.findings-acl)

Challenge:	Existing methods for dense retrieval use pseudo-relevance feedback to model dimension importance . however, they learn global transformations shared across queries and do not model dimension-aware dimension importance.
Approach:	They propose a Query-Aware Adaptive Dimension Selection framework that learns to predict per-dimension importance directly from query embedding.
Outcome:	The proposed framework improves retrieval effectiveness over the full-dimensional and PRF-based models.

Copied to clipboard

Peiyi Zhang, Richong Zhang, Zhijie Nie, Ziqiao Wang