Papers by ChengXiang Zhai
Measuring the Effect of Influential Messages on Varying Personas (2023.acl-short)
Copied to clipboard
| Challenge: | a new task estimates the response a persona might have upon seeing a news message . a first benchmark dataset is used to evaluate the performance of the proposed task . |
| Approach: | They propose a task to estimate the response a persona might have upon seeing a news message. |
| Outcome: | The proposed task estimates the response a persona might have upon seeing a news message. |
Text2Mol: Cross-Modal Molecule Retrieval with Natural Language Queries (2021.emnlp-main)
Copied to clipboard
| Challenge: | Existing databases contain tens of millions of molecules; PubChem alone has 110 million compounds. |
| Approach: | They propose a task to retrieve molecules using natural language descriptions as queries . they construct a paired dataset of molecules and their corresponding text descriptions . |
| Outcome: | The proposed approach improves results from 0.372 to 0.499 MRR. |
Language Model Pre-Training with Sparse Latent Typing (2022.emnlp-main)
Copied to clipboard
| Challenge: | Modern large-scale Pre-trained Language Models focus on text reconstruction, but have not sought to learn latent-level interpretable representations of sentences. |
| Approach: | They propose a new pre-training objective that enables the model to learn latent types . the objective allows the model a self-supervised way to extract sentence-level keywords . |
| Outcome: | The proposed model learns interpretable latent type categories without external knowledge and improves downstream tasks. |
Persona-DB: Efficient Large Language Model Personalization for Response Prediction with Collaborative Data Refinement (2025.coling-main)
Copied to clipboard
Chenkai Sun, Ke Yang, Revanth Gangi Reddy, Yi Fung, Hou Pong Chan, Kevin Small, ChengXiang Zhai, Heng Ji
| Challenge: | Existing research has focused on enhancing the retrieval stage and optimizing the representation of the database. |
| Approach: | They propose a framework to improve generalization across task contexts and collaborative refinement to bridge knowledge gaps among users. |
| Outcome: | The proposed framework improves generalization across task contexts and collaborative refinement to bridge knowledge gaps among users. |
BiasGRPO: Stabilizing Bias Mitigation in High-Variance Reward Landscapes via Group-Relative Policy Optimization (2026.findings-acl)
Copied to clipboard
| Challenge: | Recent preference-based fine-tuning methods have limited exploration in offline training . previous methods have been limited by the lack of exploration inherent in offline learning . |
| Approach: | They propose a method that normalizes rewards across a group of completed tasks to mitigate social bias in Large Language Models. |
| Outcome: | The proposed approach outperforms DPO and PPO in multiple benchmarks . it can overcome limitations of previous preference-based methods . |
Joint Biomedical Entity and Relation Extraction with Knowledge-Enhanced Collective Inference (2021.acl-long)
Copied to clipboard
| Challenge: | Existing methods for information extraction from biomedical texts do not utilize external knowledge . despite the exponential growth of biomedically published articles, many existing methods fall behind . |
| Approach: | They propose a framework that utilizes external knowledge for entity and relation extraction . KECI uses an initial span graph to construct a knowledge graph containing relevant background knowledge . |
| Outcome: | The proposed framework achieves state-of-the-art results in two biomedical datasets . it achieves 4.59% and 4.91% improvement in F1 scores over the state- of-the art methods . |
ORBIT: Cost-Effective Dataset Curation for Large Language Model Domain Adaptation with an Astronomy Case Study (2025.findings-acl)
Copied to clipboard
| Challenge: | General-purpose models lack depth for expert-level tasks because of limited domain-specific information. |
| Approach: | They propose a method for curating domain-specific datasets from noisy web sources to improve model performance. |
| Outcome: | The proposed model outperforms the baseline model on the astronomy benchmark and on the AstroBench. |
AnaDE1.0: A Novel Data Set for Benchmarking Analogy Detection and Extraction (2024.eacl-long)
Copied to clipboard
| Challenge: | Textual analogies that make comparisons between two concepts are often used for explaining complex ideas, creative writing, and scientific discovery. |
| Approach: | They propose a task that includes three synergistic tasks: detecting documents containing analogies, extracting text segments that make up the analogy, and identifying the (source and target) concepts being compared. |
| Outcome: | The proposed task performs well on all sub-tasks and smaller models perform better than non-finetuned ChatGPT, suggesting high task difficulty. |
Beyond Reactive Safety: Risk-Aware LLM Alignment via Long-Horizon Simulation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing alignment methods focus on reactive feedback, where immediate human perception is leveraged to judge sampled model responses as preference data for post-training. |
| Approach: | They propose a proof-of-concept framework that projects how model-generated advice could propagate through societal systems on a macroscopic scale over time, enabling more robust alignment. |
| Outcome: | The proposed framework achieves 20% improvement on existing safety benchmarks and an average win rate exceeding 70% against strong baselines. |
BERT might be Overkill: A Tiny but Effective Biomedical Entity Linker based on Residual Convolutional Neural Networks (2021.findings-emnlp)
Copied to clipboard
| Challenge: | Biomedical entity linking is a task of linking entities in biomedical documents to referent entities in a knowledge base. |
| Approach: | They propose an efficient convolutional neural network with residual connections for biomedical entity linking. |
| Outcome: | The proposed model achieves comparable or even better linking accuracy on five public datasets while having about 60 times fewer parameters. |
Improving Candidate Retrieval with Entity Profile Generation for Wikidata Entity Linking (2022.findings-acl)
Copied to clipboard
| Challenge: | Existing studies focus on Wikipedia-derived KBs, but there is little work on EL over Wikidata . EL systems have found applications in many tasks such as question answering . |
| Approach: | They propose a novel approach to linking entity mentions to referent entities in a knowledge base . they use a sequence-to-sequence model to generate the profile of the target entity . |
| Outcome: | The proposed approach achieves state-of-the-art results on three Wikidata-based datasets and strong performance on TACKBP-2010. |
Multi-task Learning for Multilingual Neural Machine Translation (2020.emnlp-main)
Copied to clipboard
| Challenge: | Existing multilingual neural machine translation systems rely on bitext training data, which is limited and costly to collect. |
| Approach: | They propose a multi-task learning framework that trains the model with the translation task on bitext data and two denoising tasks on monolingual data. |
| Outcome: | The proposed framework outperforms pre-training models for both NMT and cross-lingual transfer learning NLU tasks. |
ModelingAgent: Bridging LLMs and Mathematical Modeling for Real-World Challenges (2025.findings-emnlp)
Copied to clipboard
Cheng Qian, Hongyi Du, Hongru Wang, Xiusi Chen, Yuji Zhang, Avirup Sil, ChengXiang Zhai, Kathleen McKeown, Heng Ji
| Challenge: | Existing benchmarks for large language models fail to reflect real-world complexity . existing benchmarks often fail to capture real-life problems . |
| Approach: | They propose a benchmark that features real-world-inspired, open-ended problems from competitions . they propose 'ModelingBench' that supports multiple valid solutions . |
| Outcome: | The proposed framework outperforms baselines and produces well-grounded, creative solutions. |
Social Commonsense-Guided Search Query Generation for Open-Domain Knowledge-Powered Conversations (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Open-domain dialog generates search queries that help obtain relevant knowledge for holding informative conversations. |
| Approach: | They propose to integrate social commonsense reasoning into internet search queries . they use a commonsensible dialog system to establish connections related to the conversation topic . |
| Outcome: | The proposed framework overcomes limitations of existing query generation techniques based on explicit dialog information and produces more relevant, specific, and compelling queries. |
CONCRETE: Improving Cross-lingual Fact-checking with Cross-lingual Retrieval (2022.coling-1)
Copied to clipboard
| Challenge: | Existing fact-checking approaches focus on claims made in English due to data scarcity issue in other languages. |
| Approach: | They propose a fact-checking framework augmented with cross-lingual retrieval that aggregates evidence retrieved from multiple languages through a cross-linguistic retriever. |
| Outcome: | The proposed framework achieves 2.23% absolute F1 improvement over previous systems on a X-Fact dataset. |
Decoding the Silent Majority: Inducing Belief Augmented Social Graph with Large Language Model for Response Forecasting (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing approaches to forecast news media responses have limited exploration of how to best process and utilize these important features. |
| Approach: | They propose a framework that leverages a large language model to induce a belief-centered graph on top of an existent social network, along with graph-based propagation to capture social dynamics. |
| Outcome: | The proposed framework surpasses state-of-the-art in experimental evaluations for both zero-shot and supervised settings, demonstrating its effectiveness in response forecasting. |
When to Use What: An In-Depth Comparative Empirical Analysis of OpenIE Systems for Downstream Applications (2023.acl-long)
Copied to clipboard
| Challenge: | Existing consensus on which OpenIE model is best for each application is lacking . different assumptions made by different models and datasets have a statistically significant effect on performance, making it important to choose the most appropriate OpenIE system for one’s applications. |
| Approach: | They propose to use OpenIE to extract relation tuples from plain text to compare different models and training sets to find the best model for their applications. |
| Outcome: | The proposed models perform well on a Complex QA application. |
The Law of Knowledge Overshadowing: Towards Understanding, Predicting and Preventing LLM Hallucination (2025.findings-acl)
Copied to clipboard
Yuji Zhang, Sha Li, Cheng Qian, Jiateng Liu, Pengfei Yu, Chi Han, Yi R. Fung, Kathleen McKeown, ChengXiang Zhai, Manling Li, Heng Ji
| Challenge: | Hallucination is a persistent challenge in large language models where even with rigorous quality control, models often generate distorted facts. |
| Approach: | They propose a new framework to quantify factual hallucinations by modeling knowledge overshadowing. |
| Outcome: | The proposed framework improves model factuality on Overshadow (27.9%), MemoTrap (13.1%) and NQ-Swap (18.3%). |
Incorporating Task-Specific Concept Knowledge into Script Learning (2023.eacl-main)
Copied to clipboard
| Challenge: | Existing work on Goal-Oriented Scripts ignore usage context and personal preferences . proposed tasks are restrictive and rely on overly simplified assumptions . |
| Approach: | They propose a novel approach to Goal-Oriented Script Completion that uses concept prompting and script-oriented contrastive learning to improve performance. |
| Outcome: | The proposed approach improves on a WikiHow-based dataset. |
Cache-of-Thought: Master-Apprentice Framework for Cost-Effective Vision Language Model Reasoning (2025.emnlp-main)
Copied to clipboard
Mingyuan Wu, Jize Jiang, Haozhen Zheng, Meitang Li, Zhaoheng Li, Beitong Tian, Bo Chen, Yongjoo Park, Minjia Zhang, ChengXiang Zhai, Klara Nahrstedt
| Challenge: | Recent Vision Language Models (VLMs) have shown tremendous promise in a wide range of realworld applications, but their size has made at-scale deployment and operation challenging due to high consumption of cloud computing resource, high latency, and expensive API calls. |
| Approach: | They propose a master–apprentice framework for collaborative inference between large and small vision language models. |
| Outcome: | The proposed framework improves reasoning performance on widely-recognized and challenging general reasoning benchmarks and specifically boosts reasoning of apprentice VLMs by 36.6%. |
Domain Representative Keywords Selection: A Probabilistic Approach (2022.findings-acl)
Copied to clipboard
| Challenge: | a probabilistic approach to select a subset of a target domain representative keywords is crucial for many downstream tasks in natural language processing. |
| Approach: | They propose a probabilistic approach to select a subset of a target domain representative keywords from a candidate set, contrasting with a context domain. |
| Outcome: | The proposed approach provides more importance to distinctive keywords than common keywords contrasting with the context domain. |
UOUO: Uncontextualized Uncommon Objects for Measuring Knowledge Horizons of Vision Language Models (2024.emnlp-main)
Copied to clipboard
Xinyu Pi, Mingyuan Wu, Jize Jiang, Haozhen Zheng, Beitong Tian, ChengXiang Zhai, Klara Nahrstedt, Zhiting Hu
| Challenge: | Vision-Language Models (VLMs) perform on par with larger models in general domain visual grounding and question-answering benchmarks. |
| Approach: | They propose a "Uncontextualized Uncommon Objects" benchmark to evaluate their performance on common datasets. |
| Outcome: | The proposed benchmark focuses on systematically testing VLMs with both large and small parameter counts on rare and specialized objects. |