Method improves multi-hop question answering with reusable answers
DeFT-RAG: Incrementally Decomposing Multi-Hop Questions into Self-Contained Sub-Questions with Reusable Answers
Computation and LanguageArtificial IntelligenceInformation Retrieval
Summary
Answering complicated questions often needs gathering facts from many different documents and connecting them. The authors found that existing methods struggle because they either miss important information or overload the language model with many tasks. They created DeFT-RAG, which breaks hard questions into smaller, clear sub-questions and remembers past answers to avoid repeating work. This approach makes answering multi-step questions more accurate and efficient.
What this means in practice
- •For customer support teams: Improve automated systems by accurately answering complex, multi-step user inquiries using fewer model resources.
- •For business intelligence analysts: Enable more reliable document exploration by progressively breaking down complex queries with answer reuse to speed insight gathering.
Authors
Dingyi Kang, Bingzhe Li
Abstract
Retrieval-augmented generation (RAG) lets a large language model (LLM) answer questions by retrieving relevant passages from a large corpus. However, RAG struggles with multi-hop questions, which require chaining facts spread across several documents through intermediate entities that the question never mentions. Graph-based RAG methods retrieve once from the entities a question names and so rarely recover the full evidence chain. Iterative RAG methods recover it step by step, but their accuracy is limited by the many tasks they ask the LLM to perform, such as planning the sub-questions, deciding their dependencies, rewriting them as answers arrive, and keeping track of what is known. We propose DeFT-RAG, a RAG framework built on three components. First, DeFT-RAG generates one self-contained sub-question at a time from the original question and the sub-questions answered so far with their answers (QA pairs). Each sub-question names every entity it asks about, which removes the need to plan, track dependencies or rewrite, and lets the LLM focus on a single fine-grained task at a time. Second, DeFT-RAG moves the remaining LLM work outside the LLM: rule-based updates to a QA table that keeps track of what is known, and yes/no decisions made by a fine-tuned lightweight decision model. Third, DeFT-RAG caches the QA pairs and reuses an answer when a later question asks a sub-question with the same meaning. On MuSiQue, 2WikiMultiHopQA and HotpotQA, DeFT-RAG achieves the best accuracy among thirteen baselines with both gpt-4o-mini and Qwen3.6 as the backbone, while giving the LLM fewer input tokens than any other iterative method. With a QA cache shared across 1,000 MuSiQue questions, 23.1% of sub-questions are answered from the cache, which reduces LLM calls by 12.9% without lowering accuracy.