Revisit Semantic Representation and Tree Search for Similar Question Retrieval

08/22/2019

∙

by   Tong Guo, et al.

∙

0

∙

share

In this paper, we study the problem of short sentence ranking for question answering. In order to get best score for all the sentences when given a query. We compute the representation for all the sentences in advance and leverage k-d tree to accelerate the speed. The experimental results shows that our methods beat the strong baseline of BM25 on large information retrieval corpus. We will compare our experiment results to other representation-based neural rankers in the future. And we will do the experiment of speed comparison between BM25-based and our tree-based retrieval approach.