Abstract

Non-factoid question-answering (NFQA) poses a significant challenge due toits open-ended nature, diverse intents, and the need for multi-aspectreasoning, which renders conventional factoid QA approaches, includingretrieval-augmented generation (RAG), inadequate. Unlike factoid questions,non-factoid questions (NFQs) lack definitive answers and require synthesizinginformation from multiple sources across various reasoning dimensions. Toaddress these limitations, we introduce Typed-RAG, a type-aware multi-aspectdecomposition framework within the RAG paradigm for NFQA. Typed-RAG classifiesNFQs into distinct types -- such as debate, experience, and comparison -- andapplies aspect-based decomposition to refine retrieval and generationstrategies. By decomposing multi-aspect NFQs into single-aspect sub-queries andaggregating the results, Typed-RAG generates more informative and contextuallyrelevant responses. To evaluate Typed-RAG, we introduce Wiki-NFQA, a benchmarkdataset covering diverse NFQ types. Experimental results demonstrate thatTyped-RAG outperforms baselines, thereby highlighting the importance oftype-aware decomposition for effective retrieval and generation in NFQA. Ourcode and dataset are available at https://github.com/TeamNLP/Typed-RAG.