一作ACL 2026Poster· ACL Industry Track
A Novel Matching Paradigm: Unified Generative and Discriminative LLM with Prompt Compression for Relevance Learning
Guoliang Zhao · Zixin Cui · Chao Ye · Dengwu He · Fei Huang · Yubo Liu · Shuanglong Li · Tzungren Kuo · Bin Ding · Shuang Zhang · KunhongZhu · Zhi Guo · Liu Lin
摘要
The matching paradigm is fundamental to large-scale information retrieval and is widely used in industrial search and advertising systems. Existing approaches employ Large Language Models (LLMs) primarily as feature extractors, underutilizing their full modeling capabilities. To address this limi...