arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2410.03754cs.CLcs.IR
收藏

通过派生特征关联增强问答系统中的检索能力

Enhancing Retrieval in QA Systems with Derived Feature Association

University of Pennsylvania(宾夕法尼亚大学)

Keyush Shah, Abhishek Goyal, Isaac Wasserman

AI总结:

针对传统RAG朴素检索机制在主观长上下文问答任务中易漏检答案的问题,提出RAIDD扩展方案,通过在文档入库时用LLM提取摘要、示例问题等派生特征辅助检索,显著提升了长上下文问答任务性能。

摘要:

检索增强生成(RAG)已成为长上下文问答(QA)系统的标准方案。然而,典型的RAG实现依赖相当朴素的检索机制:嵌入与查询嵌入最相似的文本会被判定为最相关。这在主观问答任务中会产生问题——最相关的文本可能并不直接包含答案。本研究提出一种RAG系统的新型扩展方案,名为Retrieval from AI Derived Documents(RAIDD,即从AI派生文档中检索)。RAIDD在检索阶段充分利用大语言模型(LLM)的能力,在文档入库时从文档中提取摘要、示例问题等推断特征。我们证实,该方法能显著提升RAG系统在长上下文问答任务上的性能。

英文摘要:

Retrieval augmented generation (RAG) has become the standard in long context question answering (QA) systems. However, typical implementations of RAG rely on a rather naive retrieval mechanism, in which texts whose embeddings are most similar to that of the query are deemed most relevant. This has consequences in subjective QA tasks, where the most relevant text may not directly contain the answer. In this work, we propose a novel extension to RAG systems, which we call Retrieval from AI Derived Documents (RAIDD). RAIDD leverages the full power of the LLM in the retrieval process by deriving inferred features, such as summaries and example questions, from the documents at ingest. We demonstrate that this approach significantly improves the performance of RAG systems on long-context QA tasks.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑