Research on question retrieval method for community question answering

When the neural network model is applied to solve the question retrieval task of community question and answer, it needs a large corpus and long retrieval time. To address these problems, this paper proposes a two-stage question retrieval algorithm. In the second stage, the multi-feature fusion method is adopted to comprehensively judge the retrieved results according to the similarity of the query sentence to the candidate question sentence in lexical features and semantic features, as well as the answer quality features in the candidate answers. Experimental results ranked second with 78.3 on SemEval-2016 Task3 test set and ranked first with 48.20 on SemEval-2017 Task3 test set and and only took 500 ms to get the results from 1000 pieces of data. These results show that this algorithm can significantly improve the question retrieval effect while ensuring the retrieval efficiency.