Method and device for calculating confidence coefficient of triple in knowledge graph

Abstract

The invention discloses a method and device for calculating the confidence coefficient of a triple in a knowledge graph, and relates to the field of artificial intelligence. According to the specificimplementation scheme, the method comprises the steps of determining a to-be-searched query statement according to a target triple in a knowledge graph; obtaining a network search result correspondingto the query statement; determining a first feature corresponding to the target triple according to a network search result; determining a second feature of the target triple according to the matching degrees of the first subject, the first object and the first relationship with each piece of information in a preset knowledge graph; and performing encoding and decoding processing on the first feature and the second feature by using a preset network model, and determining the confidence coefficient of the target triple. According to the method, the accuracy of the determined confidence coefficient is high through the first feature obtained according to the network search result and the second feature obtained according to the preset knowledge graph, so that high-quality triples are mined,the method is not limited by a mining source, the coverage rate of the knowledge graph is increased, manual participation is not needed, and the mining cost of the triples is reduced.

Description

technical field [0001] The present application relates to the field of computer technology, in particular to the field of artificial intelligence, and in particular to a method and device for calculating confidence degrees of triples in knowledge graphs. Background technique [0002] The triples in the knowledge graph (including the subject, object, and the relationship between the subject and the object) play an important role in various application scenarios such as entity question answering and entity recommendation. The accuracy of triplets directly affects the accuracy of each application. At present, in order to improve the accuracy of triplets, the quality of mining sources is usually controlled, or the method of manually labeling and mining the full amount of data is used. [0003] However, due to the limited number of high-quality mining sources, limiting the mining sources will directly affect the coverage of the knowledge map, and the cost of labeling and mining ...

AI Technical Summary

Problems solved by technology

[0004] This application proposes a method and device for calculating the confidence degree of triples in knowledge graphs, which are used to solve the problem of low coverage of knowledge graphs in related technologies, by controlling the quality of mining sources or manually improving the accuracy of triples , The problem of high cost of manual labeling

Images