Question and answer system construction method based on Pipeline processing and ES storage

Abstract

The invention relates to natural language processing, and in particular relates to a question and answer system construction method based on Pipeline processing and ES storage. The method comprises the following steps: collecting and cleaning industry field data question and answer pairs, and constructing multi-dimensional semantic feature tags for questions in the question and answer pairs; storing questions in the question and answer pairs and the multi-dimensional semantic feature tags and answers correspondingly matching the questions in an ES database in batches; judging whether the input question statement contains sensitive words or not, if the input question statement contains the sensitive words, directly giving an answer, and if not, rewritting the input question statement; performing multi-dimensional semantic feature tag extraction on the rewritten question statement, converting the rewritten question statement into a query statement, and recalling candidate question and answer pairs; performing similarity scoring and rearrangement on the rewritten question statement and the questions in the candidate question and answer pairs, and recommending an optimal answer. According to the technical scheme, the defects that in the prior art, comprehensive maintenance of the question and answer pair library is lacked, and similarity judgment of input questions and candidate questions in the library is not accurate enough can be overcome.

Description

technical field [0001] The present invention relates to natural language processing, in particular to a method for constructing a question answering system based on Pipeline processing and ES storage. Background technique [0002] Question answering system is a relatively complex and difficult task in the field of natural language processing. The task is to give a question and the system can give a reasonable and accurate answer. According to the current research progress at home and abroad, and according to the source of answers, question answering systems can be divided into structured question answering systems, text based question answering systems, and FAQ question answering system based question answering systems. [0003] Based on the structured question answering system, it is to bring questions into the structured knowledge base prepared in advance to seek answers. It can parse the input natural language questions and convert the parsing results into corresponding S...

AI Technical Summary

Problems solved by technology

[0008] Aiming at the above-mentioned shortcomings existing in the prior art, the present invention provides a method for constructing a question answering system based on Pipeline processing and ES storage, which can effectively overcome the lack of comprehensive maintenance of the question and answer database, input questions and candidates in the database. The defect that the similarity judgment of the problem is not accurate enough

Images