用轻量大模型生成关键词查询，提升检索效率与泛化能力
QueStER: Query Specification for Generative keyword-based Retrieval
- 通过轻量LLM将用户查询转化为关键词，由标准检索器执行
- 在跨域测试中优于BM25，接近神经检索基线性能
- 兼顾高效性与泛化能力，适合需要快速部署的检索场景
生成式检索（GR）不同于传统索引-检索流程，它将相关性存储在模型参数中，并直接从查询生成检索线索，但存在域外表现脆弱、扩展成本高的问题。本文提出QueStER（QUEry SpecificaTion for gEnerative Keyword-Based Retrieval），通过学习生成明确的基于关键词的搜索规范，弥合生成式检索与查询重写之间的差距。给定用户查询，一个轻量级大模型生成关键词查询，由标准检索器（BM25）执行，结合了生成式查询重写的泛化优势与词法索引的效率和可扩展性。我们使用强化学习技术训练重写策略。在跨域评估中，QueStER持续优于BM25，且性能媲美神经信息检索基线，同时保持良好效率。
原文摘要 · Abstract (English)
Generative retrieval (GR) differs from the traditional index-then-retrieve pipeline by storing relevance in model parameters and generating retrieval cues directly from the query, but it can be brittle out of domain and expensive to scale. We introduce QueStER (QUEry SpecificaTion for gEnerative Keyword-Based Retrieval), which bridges GR and query reformulation by learning to generate explicit keyword-based search specifications. Given a user query, a lightweight LLM produces a keyword query that is executed by a standard retriever (BM25), combining the generalization benefits of generative query rewriting with the efficiency and scalability of lexical indexing. We train the rewriting policy with reinforcement learning techniques. Across in- and out-of-domain evaluations, QueStER consistently improves over BM25 and is competitive with neural IR baselines, while maintaining strong efficiency.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。