Cross-language knowledge graph entity alignment method based on GCN twinning network

Abstract

The invention discloses a cross-language knowledge graph entity alignment method based on a GCN twinning network. The method comprises the steps: firstly respectively extracting an adjacent matrix andan attribute information matrix for knowledge graphs of two languages, enabling the known alignment entity pairs of the knowledge graphs of the two languages to serve as positive samples, and forminga positive sample set; respectively constructing a GCN twin network for the relationship structure information and the attribute information, respectively recording the relationship structure information and the attribute information as GCN _ SE and GCN _ AE, and training the GCN twin network GCN _ SE by taking adjacent matrixes of the two knowledge maps as inputs of the GCN twin network GCN _ SE; taking the adjacency matrix and the attribute information matrix of the two knowledge maps as the input of a GCN twin network GCN _ AE, and training the GCN twin network GCN _ AE; and obtaining a potential alignment entity result based on the trained GCN twin network. According to the method, only relation structure information and attribute information between entities in the multi-language knowledge graph and part of aligned entities are needed as a training set, and multiple potential aligned entity pairs can be inferred at the same time without acquiring extra entity feature data.

Description

technical field [0001] The invention belongs to the technical field of natural language processing, and more specifically, relates to a method for aligning cross-language knowledge graph entities based on GCN twin networks. Background technique [0002] With the development of the Internet, the amount of data information has exploded, the data is redundant, and the scale is huge. In order to solve this series of problems, the knowledge graph technology aimed at describing the entities existing in the real world and the relationship between entities was born. In order to obtain a more complete knowledge map, a method of integrating multiple knowledge bases in different languages ​​can be used to obtain a multilingual knowledge map containing more information and entities. There are some known cross-language links in this kind of knowledge graph, connecting the same entity pointed to by multiple languages, that is, entity alignment. For example, there is an entity "Shanghai P...

AI Technical Summary

Problems solved by technology

Most of the traditional entity alignment methods adopt the method of predicting through the respective attributes of the entities, ignoring the correlation between the entities themselves, or not making full use of the interaction between attribute information and relationship information, or in the negative sampling stage The correct negative samples are not obtained for learning, so the overall alignment accuracy rate is not high

Images