Knowledge graph inconsistency reasoning method based on neural network

Abstract

The invention discloses a knowledge graph inconsistency reasoning method based on a neural network, and the method comprises the following steps: carrying out the learning of triad representation through a knowledge representation learning algorithm, obtaining entity representation and relation representation, calculating a representation score, and enabling the entity representation and relationrepresentation to serve as the input of the neural network, modeling the axiom through the neural network by utilizing the triad so as to learn parameters of the neural network used for representing the corresponding axiom to obtain an axiom model, obtaining an axiom prediction value of the triad by utilizing the axiom model, and judging the inconsistency of the triad and the corresponding axiom based on the representation score of the triad and the axiom prediction value. According to the method, ontology information does not need to be given, the inconsistency axiom is learned through the neural network, and whether inconsistency exists in a triple or not and whether inconsistency exists in the given axiom or not are judged through a knowledge representation learning algorithm and the neural network.

Description

technical field [0001] The invention belongs to the fields of knowledge graphs and neural networks, and in particular relates to a neural network-based inconsistency reasoning method for knowledge graphs. Background technique [0002] Knowledge graph is a knowledge system formed by structuring knowledge, which has been widely used in knowledge-driven tasks such as search engines, recommendation systems, and question answering systems. In order to efficiently store and utilize knowledge, people use manual labeling, semi-automatic or automatic methods to build large-scale knowledge graphs for open domains and vertical domains. Classical knowledge graphs, such as Wikidata, Freebase, and DBpedia, use triples to store the relationship between entities and realities. Each triple corresponds to a piece of knowledge, for example (China, the capital is, Beijing) means that the capital of China is Beijing, where "China" is called the head entity, "Beijing" is called the tail entity, ...

AI Technical Summary

Problems solved by technology

However, these methods can only judge the consistency of the triples, and cannot judge whether the axioms corresponding to the triples are consistent in a fine-grained manner. Therefore, there is an urgent need for a method to detect false knowledge by detecting whether a certain axiom of the triples is consistent. for error correction

Images