Knowledge graph completion method based on graph perception tensor decomposition

Abstract

The invention discloses a knowledge graph completion method based on graph perception tensor decomposition, and the method comprises the following steps: extracting the representation information of triple data (es, r, eo) from a graph neural network, and constructing a graph coding model with an entity and a relation, i.e. G = (V, E); constructing a three-order tensor decomposition model for thetwo-dimensional representation information of the graphic coding model through a Tucker decomposition method; namely, the three-order tensor decomposition model takes the maximum probability of prediction (es, r,) as the probability output that a triple is true, knowledge graph complementation is achieved, the problems that in an existing knowledge graph library, the relation between data is speculated, and the implicit connection relation between entities is difficult to mine are solved. High-precision completion of a large-scale knowledge graph data set is realized.

Description

technical field [0001] The invention relates to the field of knowledge graphs, in particular to a knowledge graph completion method based on graph-aware tensor decomposition. Background technique [0002] On May 17, 2012, Google first proposed the concept of Knowledge Graph, which is used to refer to a knowledge base that improves search engine performance and improves user search experience. The advent of the big data era has led to the enrichment of knowledge graph databases. Commonly used public knowledge graphs include FreeBase, OpenKG, Yago, DBpedia, etc. [0003] Knowledge graph is a kind of network-shaped visual data, which is used to describe knowledge resources and their carriers, mine, analyze, construct, draw and display knowledge and their interrelationships. The knowledge map is expressed in the form of triples (head, relation, tail). Head represents the head entity, and tail represents the tail entity. In the figure, they are collectively referred to as entiti...

AI Technical Summary

Problems solved by technology

[0007] Aiming at the technical problems existing in the prior art, the present invention provides a knowledge graph completion method based on graph-aware tensor decomposition, which can combine graph neural network and knowledge graph completion through tensor decomposition, thereby solving existing The problem of inferring the relationship between data in the knowledge graph library and mining the hidden connection relationship between entities is difficult, so as to complete the knowledge graph and realize the high-precision completion of large-scale knowledge graph data

Images