Urban domain knowledge graph credibility evaluation method and system

Abstract

The invention discloses an urban domain knowledge graph credibility evaluation method and system. The method comprises the following steps: S100, obtaining credible evidence of a knowledge obtaining stage, a knowledge reasoning stage, and a knowledge verification stage from a domain knowledge credible evidence model; S200, on the basis of the obtained credible evidence and a calculation method of credible attribute characteristic values in a domain knowledge credible attribute model, credible attribute characteristic values of knowledge entities and a knowledge model are calculated; S300, calculating the credibility of the knowledge entities based on the credible attribute characteristic values of the knowledge entities, or calculating the credibility of the knowledge model based on the credible attribute characteristic values of the knowledge model; S400, dividing credibility levels of the knowledge entities or the knowledge model. According to the method the system, the credibility of knowledge or knowledge models can be quantified, so that the quality of the urban domain knowledge models is guaranteed, and the knowledge requirements of different users in different domains are met.

Description

technical field [0001] The present invention relates to the field of knowledge graphs, in particular to a method and system for credible evaluation of knowledge graphs in the city domain. Background technique [0002] Knowledge graphs describe facts in the real world through knowledge triples. In the process of building knowledge graphs, it is inevitable to introduce noise. The existing knowledge graph systems lack the credibility evaluation and guarantee of knowledge, which will give Knowledge graph applications bring potential damage. There are diverse application scenarios in cities that have significantly different requirements for knowledge credibility, such as city attraction recommendation, urban development decision-making, and urban epidemic monitoring. In order to provide credible knowledge services for diverse application scenarios in cities, knowledge The evaluation and guarantee of the credibility of knowledge in graphs has become a key issue that needs to be s...

AI Technical Summary

Problems solved by technology

In the existing technology, some methods use representation learning to represent knowledge triples as low-dimensional vectors, and detect noise in knowledge graphs through vector distance. These methods can denoise knowledge graphs to a certain extent, but there is no Quantify the credibility of knowledge

Images