Knowledge graph entity alignment method based on similarity relationship probability reasoning

Abstract

The invention belongs to the field of artificial intelligence and natural language processing, and particularly relates to a knowledge graph entity alignment method and device based on probabilistic reasoning, and the method comprises the steps: respectively obtaining entity alignment seeds used for training and a relation connected with the entity alignment seeds across knowledge graphs; the entities and the relationships are put into a knowledge graph representation learning algorithm for training; based on the relation vector obtained by training, calculating the similarity of the relation by using a vector similarity algorithm, and extracting a relation pair with relatively high similarity; on the basis of the entity alignment seeds and the relation pairs obtained through calculation, the alignment possibility of non-aligned entities can be iteratively calculated through probabilistic reasoning, and the entity alignment result is labeled; according to the method, the knowledge graph representation learning algorithm and the novel knowledge graph probabilistic reasoning algorithm are utilized, so that the entity alignment effect of the knowledge graph can be effectively improved.

Description

technical field [0001] The invention belongs to the field of artificial intelligence and natural language processing, and in particular relates to a knowledge map entity alignment method based on similarity relationship and probability reasoning. Background technique [0002] With the rapid development of the Internet, both the data scale and the data types in today's Internet show an exponential growth phenomenon. In order to utilize a large amount of unstructured data in the Internet (such as human natural language) for systematic storage, management and application, people extract knowledge from the Internet to build a structured knowledge map. Knowledge graph is the basic core technology of artificial intelligence research and intelligent information service. It can endow intelligent agents with the ability of precise query, attempt to understand and logical reasoning. It is widely used in knowledge-driven tasks such as semantic search, knowledge question answering and p...

AI Technical Summary

Problems solved by technology

Existing knowledge map entity alignment works are based on traditional probabilistic reasoning systems and representation learning models, both of which have their own advantages and disadvantages. For example, traditional probabilistic reasoning systems need to use the name and text information in the knowledge map. If two knowledge graphs use different symbols to represent the same entity when they are constructed, then this probabilistic reasoning system cannot run

Models based on representation learning map entities and relationships into vector spaces, and align entities through distance searches in vector spaces. Since all similar entities may gather together in high-dimensional vector spaces, there is no obvious The distinction often leads to the problem of poor alignment

Images