Knowledge graph fusion method based on entity alignment

Abstract

The invention discloses a knowledge graph fusion method based on entity alignment. The knowledge graph fusion method comprises the following steps: acquiring data of two knowledge graphs; learning thestructure vector of the entity by using a graph convolution network, and expressing the name of the entity as a word vector; calculating a comprehensive distance between the entities to represent a similarity degree between the entities; performing entity identification alignment by adopting an iterative training framework based on course learning; and according to an entity alignment result, fusing the two knowledge maps into one knowledge map. According to the method, an entity alignment basic framework fusing structural features and entity name features is designed; an iterative training method based on curriculum learning is designed, training data are amplified from easy to difficult, and a word shift distance model is adopted to reorder preorder alignment results, so that entity name information is fully mined, and knowledge graph fusion is more accurate and comprehensive.

Description

technical field [0001] The invention belongs to the field of knowledge graph generation and fusion, and in particular relates to a knowledge graph fusion method based on entity alignment. Background technique [0002] In recent years, a large number of knowledge graphs (knowledge graph, KG) have emerged, such as YAGO, DBpedia, NELL, and Chinese CN-DBpedia, Zhishi.me, etc. These large-scale knowledge graphs play an important role in intelligent services such as question answering systems and personalized recommendations. In addition, in order to meet the needs of specific domains, more and more domain knowledge graphs, such as medical knowledge graphs, have been derived. In the process of building a knowledge graph, it is inevitable to make a trade-off between coverage and accuracy. However, any knowledge map cannot be complete or completely correct. [0003] In order to improve the coverage and accuracy of the knowledge graph, a feasible method is to introduce relevant kn...

AI Technical Summary

Problems solved by technology

To solve this problem, some methods propose to use iterative training (IT) to select high-confidence entity pairs from the test set results for the next round of training, but there are problems such as easy introduction of error samples and low efficiency.

In addition, on datasets with real-world distributions, these iterative training frameworks can only introduce a small number of high-confidence entity pairs, which cannot bring significant performance improvements.

Images