Huimin Xu

Scaling up Open Tagging from Tens to Thousands: Comprehension Empowered Attribute Value Extraction from Product Title

2019

Supplementing product information by extracting attribute values from title is a crucial task in e-Commerce domain. Previous studies treat each attribute only as an entity type and build one set of NER tags (e.g., BIO) …

Don't Forget Your Reward Values: Language Model Alignment via Value-based Calibration

2024 · arXiv (Cornell University)

While Reinforcement Learning from Human Feedback (RLHF) significantly enhances the generation quality of Large Language Models (LLMs), recent studies have raised concerns regarding the complexity and instability associated with the Proximal Policy Optimization (PPO) algorithm, …

Negation Triplet Extraction with Syntactic Dependency and Semantic Consistency

2024 · arXiv (Cornell University)

Previous works of negation understanding mainly focus on negation cue detection and scope resolution, without identifying negation subject which is also significant to the downstream tasks. In this paper, we propose a new negation triplet …

Auto-TA: Towards Scalable Automated Thematic Analysis (TA) via Multi-Agent Large Language Models with Reinforcement Learning

2025 · arXiv (Cornell University)

Congenital heart disease (CHD) presents complex, lifelong challenges often underrepresented in traditional clinical metrics. While unstructured narratives offer rich insights into patient and caregiver experiences, manual thematic analysis (TA) remains labor-intensive and unscalable. We propose …