Answer sorting method and device and storage medium

Abstract

Embodiments of the invention disclose an answer sorting method and device, and a storage medium. The method comprises the steps of obtaining a plurality of candidate answers corresponding to a question; according to the matching information between the question and each candidate answer, calculating a correctness score that each candidate answer is a correct answer; according to the correctness score of each candidate answer, pre-sequencing the candidate answers to obtain a pre-sequencing queue; determining a target candidate answer which currently needs to participate in reordering from the pre-ordering queue according to the ordering sequence of each candidate answer in the pre-ordering queue; extracting target evidence information from the background information of the historical candidate answers, and obtaining target correctness scores of the target candidate answers according to the target evidence information and the questions; inserting the target candidate answers into corresponding positions in the target sorting queue according to the target correctness scores; and when all candidate answers in the pre-sorting queue are inserted into the target sorting queue, outputtingthe target sorting queue. According to the scheme, the accuracy of judging the candidate answers can be improved.

Description

technical field [0001] The embodiments of the present application relate to the technical field of the Internet, and in particular to an answer sorting method, device and storage medium. Background technique [0002] In a search system, multiple answers are generally given to the user's input question for the user to choose, which is referred to as answer selection for short. Answer selection is a sorting process that ranks correct candidate answers above incorrect ones. At present, answer selection is mainly based on two types of models. One is based on the similarity matching between questions and candidate answers. Questions are the only basis for judging candidate answers, and other additional information is not considered. One is that in addition to using the question as a basis, it also tries to extract some additional information from an additional corpus or knowledge base as the basis for selecting the correct answer. For example, using a deep network structure suc...

AI Technical Summary

Problems solved by technology

The additional information used includes the text related to the question searched by the search system, and the triples extracted from the knowledge base through entity links. However, this method of extracting additional information requires the use of third-party tools, so it takes a long time and expensive

The accuracy of judging candidate answers in these two ways is not high

Images