Method, device, medium, equipment and system for optimizing edge weights in knowledge map

Abstract

The invention relates to a method for optimizing edge weights in a knowledge map, comprising the following steps: defining the knowledge map, wherein the knowledge map comprises nodes, directed edgesamong the nodes and the original edge weights of the directed edges; receiving query information from a user; searching candidate answers corresponding to the query information, sorting the candidateanswers by utilizing the knowledge map, and pushing to the user; receiving feedback information on the sorted candidate answers from the user; constructing a signomial geometric programming problem, wherein a constraint function of the signomial geometric programming problem is set based on the feedback information, and a target function of the signomial geometric programming problem is an edge weight optimizing function; and solving the signomial geometric programming problem to obtain the optimized edge weights. The method provided by the invention converts an edge weight optimization problem of the knowledge map into one signomial geometric programming problem by utilizing the feedback information of the user and can efficiently and automatically improve accuracy of the edge weights andquality of the knowledge map. The invention also relates to a device, a medium, equipment and a system for optimizing the edge weights in the knowledge map.

Description

technical field

[0001] The present invention relates to the field of knowledge map optimization, in particular to a method, device, medium, equipment and system for optimizing edge weights in a knowledge map. Background technique

[0002] In modern society, knowledge graphs have been widely used in various applications, such as question answering (Q&A) systems, recommender systems, web search engines, and precision medicine, etc. For example, computing the similarity between questions and help documents based on knowledge graphs has been proven to be an effective way to find the best document answers.

[0003] In the knowledge graph, an edge represents the association information between two entity nodes, and the strength of the association is usually represented by the weight of the edge. Edge weight information can show high value in knowledge graph applications. For example, in a technical support question answering system application, if the problem of "computer crash"...

Images