Question-to-Question Retrieval for Hallucination-Free Knowledge Access: An Approach for Wikipedia and Wikidata Question Answering
This addresses hallucination in knowledge base question answering for applications like Wikipedia and Wikidata, though it appears incremental as it adapts existing retrieval techniques to a new question-based paradigm.
The paper tackles the problem of hallucination in knowledge base question answering by introducing a question-to-question retrieval approach that matches user queries against a pre-generated set of questions for each content unit, achieving high cosine similarity (>0.9) for relevant pairs and enabling precise, direct content retrieval without answer generation.
This paper introduces an approach to question answering over knowledge bases like Wikipedia and Wikidata by performing "question-to-question" matching and retrieval from a dense vector embedding store. Instead of embedding document content, we generate a comprehensive set of questions for each logical content unit using an instruction-tuned LLM. These questions are vector-embedded and stored, mapping to the corresponding content. Vector embedding of user queries are then matched against this question vector store. The highest similarity score leads to direct retrieval of the associated article content, eliminating the need for answer generation. Our method achieves high cosine similarity ( > 0.9 ) for relevant question pairs, enabling highly precise retrieval. This approach offers several advantages including computational efficiency, rapid response times, and increased scalability. We demonstrate its effectiveness on Wikipedia and Wikidata, including multimedia content through structured fact retrieval from Wikidata, opening up new pathways for multimodal question answering.