Visconde: Multi-document QA with GPT-3 and Neural Reranking
This addresses the challenge of multi-document question answering for AI systems, though it is incremental as it builds on existing retrieval and LLM methods.
The paper tackled the problem of answering questions requiring evidence from multiple documents by proposing Visconde, a three-step pipeline using LLMs for decomposition and aggregation with a search engine for retrieval, achieving human-level performance on three datasets when relevant passages are provided.
This paper proposes a question-answering system that can answer questions whose supporting evidence is spread over multiple (potentially long) documents. The system, called Visconde, uses a three-step pipeline to perform the task: decompose, retrieve, and aggregate. The first step decomposes the question into simpler questions using a few-shot large language model (LLM). Then, a state-of-the-art search engine is used to retrieve candidate passages from a large collection for each decomposed question. In the final step, we use the LLM in a few-shot setting to aggregate the contents of the passages into the final answer. The system is evaluated on three datasets: IIRC, Qasper, and StrategyQA. Results suggest that current retrievers are the main bottleneck and that readers are already performing at the human level as long as relevant passages are provided. The system is also shown to be more effective when the model is induced to give explanations before answering a question. Code is available at \url{https://github.com/neuralmind-ai/visconde}.