Papers for

policy analysis teams

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

TimelyRAG improves question answering on changing documents with overlapping updates

TimelyRAG: Semantic-Temporal Hybrid Retrieval for Time-Critical Question Answering in Overlapping-Evolving Documents

Abstract: Although large language models (LLMs) and retrieval-augmented generation (RAG) have advanced open-domain question answering (QA), they remain unreliable when documents evolve through amendments. Existing time-sensitive retrieval methods address only the disjoint-evolving environment, where each update is an independent snapshot. However, laws, policies, and regulations often operate in overlapping-evolving environments, where amendments override earlier clauses while preserving most content, creating strong semantic overlap across versions. We propose TimelyRAG, a retriever-agnostic framework that incorporates temporal distance into ranking to align queries with version-appropriate documents. We also introduce TimelyQABench, the first benchmark for regulation-heavy domains with overlapping-evolving challenges. Experiments show consistent gains, up to +28.6% in nDCG@10, highlighting the importance of temporal reasoning for reliable QA over evolving documents. All resources are available at https://github.com/kaist-dmlab/TimelyRAG.

Thu 10 SeptInformation Retrieval
The gist
Information like laws and policies often change over time by updating existing documents rather than replacing them entirely. This makes it hard for computer systems to find the right and most recent version to answer questions accurately. The authors developed TimelyRAG, a method that considers both the meaning and timing of updates to better match questions with the correct document version. They also created a benchmark called TimelyQABench to test these challenges in regulation-heavy texts, showing that accounting for document timing greatly improves question answering accuracy.
Open → 2609.11572v1