Expert domain knowledge graph query method based on natural language questions

Abstract

The invention relates to an expert domain knowledge graph query method based on natural language questions, and the method comprises the steps: carrying out the word segmentation of an obtained queryquestion through a HanLP Chinese language processing package, obtaining a feature word of the query question, and obtaining a keyword and a part-of-speech; replacing the keywords with the part-of-speech of the corresponding keywords, and performing abstract representation of the query questions; carrying out question classification on the converted abstracted sentences based on a naive Bayes classifier, firstly defining a query question template set, training the question template set by utilizing the naive Bayes classifier, then carrying out probability prediction on the input question sentences, and returning defined question category labels so as to match a question template; restoring the keywords before abstraction by using the matched question template, and returning classification labels and keyword entity names of the question template; and generating a Cypher statement according to the classification label and the keyword entity name to carry out Neo4j query, and returning a result to the user.

Description

technical field [0001] The present invention relates to natural language knowledge map technology, in particular to an expert field knowledge map query method based on natural language questions. Background technique [0002] The essence of knowledge graph is a large-scale semantic network. Since Google proposed knowledge graph in 2012, knowledge graph has been developing rapidly, and its application is no longer limited to the category of "semantic network". With the advent of the era of big data, massive data, powerful computing power, and crowd intelligence computing have emerged. Knowledge graphs have brought about qualitative changes from the scale of knowledge through the construction of large-scale, high-quality knowledge bases. Especially in semantic search, intelligent question answering, data analysis, natural language processing, visual understanding, and Internet of Things devices, etc., it has shown increasing value. In the Internet era, there is an urgent need...

AI Technical Summary

Problems solved by technology

However, in practical applications, the knowledge map query based on the Neo4j graph database is carried out through Neo4j's unique Cypher query statement, which has strong professional skills and cannot satisfy the diversification of user query input, and cannot accurately understand the query of user natural language questions. intent, low error tolerance

Images