Papers by Yumeng Zhang

11 papers
KnowledgeBerg: Evaluating Systematic Knowledge Coverage and Compositional Reasoning in Large Language Models (2026.findings-acl)

Copied to clipboard

Challenge: Existing LLMs lack systematic coverage of a bounded knowledge universe and compositional set-based reasoning over that universe.
Approach: They propose a benchmark for multiple-choice questions based on 1,183 enumeration seeds . they use knowledge width, cardinality of required universe, reasoning depth to formalize the challenge .
Outcome: The proposed benchmarks achieve only 5.26–36.88 F1 on universe enumeration and 16.00–44.19 accuracy on knowledge-grounded reasoning.
GeoLaux: A Benchmark for Evaluating MLLMs’ Geometry Performance on Long-Step Problems Requiring Auxiliary Lines (2026.acl-long)

Copied to clipboard

Challenge: Existing benchmarks for Geometry problem solving lack fine-grained evaluation for long-step problems necessitating auxiliary line construction.
Approach: They present a fine-grained annotated dataset with long-step reasoning and auxiliary line construction that provides a detailed evaluation of 23 leading MLLMs.
Outcome: The proposed model performs significantly worse on long-step problems than short-step ones, with 18 models showing a performance drop of over 50%.
Towards Better Utilization of Multi-Reference Training Data for Chinese Grammatical Error Correction (2024.findings-acl)

Copied to clipboard

Challenge: a high proportion of Chinese training data is multi-referenced for the grammatical error correction task . however, there are many ways to correct an erroneous input sentence . a systematic study on multi-referencing training data has been proposed .
Approach: They propose two new approaches and a simple two-stage training strategy to better utilize multi-reference training data.
Outcome: The proposed methods show that Chinese training data contain multiple references.
DISC: Plug-and-Play Decoding Intervention with Similarity of Characters for Chinese Spelling Check (2025.acl-long)

Copied to clipboard

Challenge: Chinese spelling check (CSC) tasks require that incorrect characters are usually similar to the correct ones in either phonetics or glyph.
Approach: They propose a plug-and-play decoding intervention with similarity of characters module for Chinese spelling check (CSC) they propose to incorporate phonetic and glyph similarities only during the inference phase.
Outcome: The proposed method significantly improves Chinese spelling check models on benchmarks and on benchmark datasets.
CLEAN–EVAL: Clean Evaluation on Contaminated Large Language Models (2024.findings-naacl)

Copied to clipboard

Challenge: Existing methods to evaluate large language models are prone to data contamination.
Approach: They propose a method which parses contaminated data and back-translates it into a candidate set.
Outcome: The proposed method reduces data contamination and evaluates the LLMs more cleanly.
Unveiling the Generalization Power of Fine-Tuned Large Language Models (2024.naacl-long)

Copied to clipboard

Challenge: Large Language Models (LLMs) have demonstrated exceptional multitasking abilities, but the comprehensive effects of fine-tuning on the LLMs’ generalization ability are not fully understood.
Approach: They conduct extensive experiments across five distinct language tasks on different datasets to investigate whether fine-tuning affects the generalization ability intrinsic to LLMs.
Outcome: The proposed model can generalize to different domains and tasks by integrating the in-context learning strategy during fine-tuning on generation tasks.
Stephanie: Step-by-Step Dialogues for Mimicking Human Interactions in Social Conversations (2025.findings-naacl)

Copied to clipboard

Challenge: a new paradigm for dialogue systems is being developed to mimic human interactions . the current single-step dialogue paradigm lacks the depth and fluidity of human interactions.
Approach: They propose a step-by-step dialogue paradigm that mimics human interactions . they use a dataset to fine-tune existing language models .
Outcome: The proposed system mimics the dynamic nature of human conversations . it is compared with existing paradigms and will be released later this year .
ERCThinker: Fast-Slow Thinking for Emotion Recognition in Conversation (2026.acl-long)

Copied to clipboard

Challenge: Existing methods for ERC lack interpretability and shallow semantics capture deep semantics.
Approach: They propose a Fast-Slow thinking framework for Emotion Recognition in Conversation . they use fine-grained emotion reasoning chains to capture deep semantics .
Outcome: The proposed framework achieves state-of-the-art in explanation and judgment on a benchmark dataset.
Improving Seq2Seq Grammatical Error Correction via Decoding Interventions (2023.findings-emnlp)

Copied to clipboard

Challenge: Existing approaches to grammatical error correction (GEC) are sequence-to-sequence and sequence-edit.
Approach: They propose a unified decoding intervention framework that employs an external critic to assess the appropriateness of the token to be generated incrementally.
Outcome: The proposed framework outperforms baselines and state-of-the-art methods on English and Chinese datasets.
LaERC-S: Improving LLM-based Emotion Recognition in Conversation with Speaker Characteristics (2025.coling-main)

Copied to clipboard

Challenge: Emotion recognition in conversation (ERC) is a task of discerning human emotions for each utterance within a conversation.
Approach: They propose a framework that uses large language models to analyze speaker characteristics . they use two-stage learning to make the models reason speaker characteristics and track emotion of the speaker .
Outcome: The proposed framework outperforms existing methods on three benchmark datasets.
JoPR: Joint Emotion Perception and Reasoning for Conversational Emotion Recognition (2026.acl-long)

Copied to clipboard

Challenge: Existing methods for ERC lack human-like emotion reasoning and discrimination between similar emotions.
Approach: They propose a multi-dimension curriculum with long CoT fine-tuning to clone human-like emotion reasoning for conversational emotion recognition.
Outcome: The proposed model outperforms existing methods on three widely used datasets and shows that it is more intuitive and more accurate.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations