Coarse-grained decomposition and fine-grained interaction for multi-hop question answering

In recent years, question answering (QA) and reading comprehension (RC) has attracted much attention, most research on QA focused multi-hop task which requires connecting multiple pieces of evidence scattered in a long context to answer the question. The key is semantic feature interaction between documents questions, widely processed by Bi-directional Attention Flow (Bi-DAF), but Bi-DAF generally captures only surface semantics words complex fails capture implied intermediate answers, as well ignoring parts contexts related failing extract important documents. this paper, we propose new model architecture for applying two completion strategies:(1) Coarse-Grained Decomposition (CGDe) strategy introduced decompose questions into simple ones without any additional annotations; (2) Fine-Grained Interaction (FGIn) explicitly represent each word more comprehensive accurate sentences inference path. above strategies are combined tested SQuAD HotpotQA datasets, experimental results show that our method outperforms state-of-the-art baselines.