Knowledge graph embedding-based character relationship mining method and device and medium

Abstract

The invention provides a method and a device for mining a character relationship based on knowledge graph embedding and a medium, which can better express a mutual relationship between an entity space and a relationship space in a semantic space so as to more accurately mine a hidden mutual relationship and a suspicious relationship between characters. The method compriss the following steps: extracting an original entity and an original relationship from a knowledge graph with a character relationship, and carrying out interactive embedding on the original relationship and the original entity to obtain an interactive unit. A prediction model is constructed and trained, the prediction model comprises a feedforward neural network layer, a convolutional neural network layer and a scoring layer, after the interaction unit is subjected to embedding representation, the feedforward neural network layer and the convolutional neural network layer are input respectively, predicted entity embedding is obtained, the predicted entity embedding is input into the scoring layer, and obtaining an entity embedding score predicted by the scoring layer; and outputting the score embedded by the predicted entity with the highest score as the mined character relationship.

Description

technical field [0001] The present invention relates to the field of knowledge graphs, in particular to a method, device and medium for mining character relationships based on knowledge graph embedding. Background technique [0002] Knowledge graphs are playing an important role in detecting and preventing crime. The knowledge graph contains a large number of existing facts, and each fact forms a triplet (h, r, t), including the head entity h, tail entity t and relation r. Although a large number of facts have been discovered in knowledge graphs, they are still incomplete and limited to our cognition. In order to solve this problem, the task of knowledge graph completion is to predict a speculative fact, that is, to predict a speculative element in a triple, so that the technology can further expand the knowledge graph on the basis of existing facts, so that it can be used to reveal Hidden interrelationships and suspicious relationships between characters to discover and p...

AI Technical Summary

Problems solved by technology

Due to the lack of adequate relational embedding learning, these knowledge graph embedding methods are difficult to extract semantic features in the relational space

Images