Papers by Kyumin Lee
Empowering Large Language Models for Textual Data Augmentation (2024.findings-acl)
Copied to clipboard
| Challenge: | True. True. False |
| Approach: | False slants are proposed to generate a large pool of augmentation instructions and select the most suitable task-informed instructions. |
| Outcome: | False omissions: the proposed approach consistently generates augmented data with better quality compared to non-LLM and LLM-based data augmentation methods. |
Piece of Table: A Divide-and-Conquer Approach for Selecting Subtables in Table Question Answering (2026.acl-long)
Copied to clipboard
| Challenge: | Existing approaches to QA tables rely on independent row or column selection, fail to capture cross-row and cross-column dependencies, or attempt global reasoning. |
| Approach: | They propose a divide-and-conquer subtable selection framework that aggregates local evidence without requiring explicit global reasoning. |
| Outcome: | The proposed framework outperforms previous approaches to table QA in the noisy context. |
ALERT: An LLM-powered Benchmark for Automatic Evaluation of Recommendation Explanations (2025.naacl-long)
Copied to clipboard
Yichuan Li, Xinyang Zhang, Chenwei Zhang, Mao Li, Tianyi Liu, Pei Chen, Yifan Gao, Kyumin Lee, Kaize Ding, Zhengyang Wang, Zhihan Zhang, Jingbo Shang, Xian Li, Trishul Chilimbi
| Challenge: | Existing benchmarks for recommendation explanation evaluation lack item diversity and user preferences data. |
| Approach: | They propose a model-agnostic recommendation explanation evaluation benchmark based on Amazon e-commerce categories with implicit preferences . they propose two novel automatic evaluators that enable scalable and human-preference aligned evaluation of explanations . |
| Outcome: | The proposed model-agnostic evaluation benchmark outperforms existing methods in a variety of domains. |
ToxiPrompt: A Two-Stage Red-Teaming Approach for Balancing Adversarial Prompt Diversity and Response Toxicity (2026.eacl-long)
Copied to clipboard
| Challenge: | Large language models (LLMs) pose safety risks, but current redteaming methods rely on human testers manually designing adversarial prompts. |
| Approach: | They propose a red-teaming method that generates adversarial prompts to elicit unsafe behavior of target LLMs. |
| Outcome: | The proposed approach outperforms state-of-the-art methods in diversity and toxicity . it performs well for multiple instruction-tuned target LLMs without re-tuning . |
Hierarchical Evidence Set Modeling for Automated Fact Extraction and Verification (2020.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for fact extraction and verification combine all evidence sentences to produce redundant information. |
| Approach: | They propose a framework to extract evidence sets and verify a claim to be supported, refuted or not enough info . they propose to encode and attend the claim and evidence sets at different levels of hierarchy . |
| Outcome: | The proposed framework outperforms 7 state-of-the-art methods for fact extraction and verification. |
Hierarchical Multi-head Attentive Network for Evidence-aware Fake News Detection (2021.eacl-main)
Copied to clipboard
| Challenge: | Existing methods to fact-check information focus on word-level attention or evidence-level focus, which may result in suboptimal performance. |
| Approach: | They propose a Hierarchical Multi-head Attentive Network to fact-check textual claims using word-level attention and document-level focus. |
| Outcome: | The proposed model outperforms state-of-the-art methods on two real-word datasets. Improvements over baselines are from 6% to 18%. |
KEPLET: Knowledge-Enhanced Pretrained Language Model with Topic Entity Awareness (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Pre-trained language models (PLMs) have shown their superiority by pre-training on unstructured text corpus and then fine-tuning on downstream tasks. |
| Approach: | They propose a Knowledge-Enhanced Pre-trained LanguagE model with Topic entity awareness that incorporates the interactions between tokens and mentioned entities in pre-training. |
| Outcome: | The proposed model incorporates the interactions between tokens and mentioned entities in pre-training and is more effective on entity-centric tasks. |
GRENADE: Graph-Centric Language Model for Self-Supervised Representation Learning on Text-Attributed Graphs (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for self-supervised representation learning on text-attributed graphs lack the full extent of structural context information or rely on task-specific training labels. |
| Approach: | They propose a Graph-Centric Language model that harnesses the synergy of pre-trained language model and graph neural network to optimize with graph-centric contrastive learning and graph-centered knowledge alignment. |
| Outcome: | The proposed model captures informative textual semantics as well as structural context information on text-attributed graphs. |
FaithfulPersona: Balancing Faithfulness and Personalization in Code Explanations through Self-Critique (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for generating faithful code explanations face challenges balancing faithfulness to the original code and personalization for diverse user needs. |
| Approach: | They propose a benchmark and method for generating faithful personalized code explanations using code samples and user profiles. |
| Outcome: | The proposed method achieves 3.7% improvement in Pass@5 compared to the strong baseline method, Self-Consistency, while maintaining high personalization with a 61.08% win rate in the LLM-as-a-Judge evaluation. |
Let’s Ask GNN: Empowering Large Language Model for Graph In-Context Learning (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Textual Attributed Graphs (TAGs) are crucial for modeling complex real-world systems, yet leveraging large language models (LLMs) for TAGs presents unique challenges due to the gap between sequential text processing and graph-structured data. |
| Approach: | They propose a novel approach that leverages In-Context Learning to integrate graph data and task-specific information into large language models (LLMs) they employ a Graph Neural Network-powered structure-enhanced retriever to select labeled nodes across graphs, incorporating complex graph structures and their supervision signals. |
| Outcome: | Experiments on three tasks and seven LLMs show that AskGNN performs better than existing methods. |
HABERTOR: An Efficient and Effective Deep Hatespeech Detector (2020.emnlp-main)
Copied to clipboard
| Challenge: | HABERTOR model is a highly efficient and effective alternative to BERT for the hatespeech classification task. |
| Approach: | They propose to modify BERT's HABERTOR model to generate its own vocabularies and pre-trained it using the largest scale hatespeech dataset. |
| Outcome: | The proposed model is faster, more efficient and more robust than existing methods for hatespeech classification. |
StepER: Step-wise Knowledge Distillation for Enhancing Reasoning Ability in Multi-Step Retrieval-Augmented Language Models (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing knowledge distillation methods overlook the need for different reasoning abilities at different steps, hindering transfer in multi-step retrieval-augmented frameworks. |
| Approach: | They propose a method that uses step-wise supervision to align with evolving information and reasoning demands across stages. |
| Outcome: | The proposed method outperforms previous methods on multi-hop QA benchmarks with an 8B model achieving performance comparable to a 70B teacher model. |
Where Are the Facts? Searching for Fact-checked Information to Alleviate the Spread of Fake News (2020.emnlp-main)
Copied to clipboard
| Challenge: | fabricated stories and hoaxes are still pervading our cyberspace. |
| Approach: | They propose a framework to search for fact-checking articles that address the content of an original tweet that may contain misinformation posted by online users. |
| Outcome: | The proposed framework can detect and disseminate fake news on real-world datasets and warn fake news posters and online users about misinformation. |