Learning to Retrieve Reasoning Paths over Wikipedia Graph for Question Answering
arXiv:1911.10470
Abstract
Answering questions that require multi-hop reasoning at web-scale necessitates retrieving multiple evidence documents, one of which often has little lexical or semantic relationship to the question. This paper introduces a new graph-based recurrent retrieval approach that learns to retrieve reasoning paths over the Wikipedia graph to answer multi-hop open-domain questions. Our retriever model trains a recurrent neural network that learns to sequentially retrieve evidence paragraphs in the reasoning path by conditioning on the previously retrieved documents. Our reader model ranks the reasoning paths and extracts the answer span included in the best reasoning path. Experimental results show state-of-the-art results in three open-domain QA datasets, showcasing the effectiveness and robustness of our method. Notably, our method achieves significant improvement in HotpotQA, outperforming the previous best model by more than 14 points.
Published as a conference paper at ICLR 2020. Code is available at https://github.com/AkariAsai/learning_to_retrieve_reasoning_paths
References in corpus (5)
- Dynamically Fused Graph Network for Multi-hop Reasoning
- PullNet: Open Domain Question Answering with Iterative Retrieval on Knowledge Bases and Text
- Cognitive Graph for Multi-Hop Reading Comprehension at Scale
- Real-Time Open-Domain Question Answering with Dense-Sparse Phrase Index
- Answering Complex Open-domain Questions Through Iterative Query Generation