Knowledge representation method based on uncertainty ontology

Abstract

The invention provides an uncertainty ontology-based knowledge representation method. The uncertainty ontology-based knowledge representation method comprises the following steps: (1) obtaining entities, types, relationships and triples with uncertainty information from an ontology knowledge graph with uncertainty information; (2) initializing representation vectors of instances, types and relations; (3) constructing an energy equation of the uncertain triple based on a distance function of the uncertain ontology; (4) according to the distance function, constructing a loss function based on amean square error, and learning representation of entities and relationships by minimizing the loss function. The method has the advantages that uncertainty information can be reserved while knowledgerepresentation is learned.

Description

technical field [0001] The present invention relates to the technical field of natural language processing and knowledge representation learning, and more specifically, to a knowledge representation method based on an uncertainty ontology. Background technique [0002] Knowledge graph is a method of representing knowledge by using entities and relationships. Currently, there are many large-scale knowledge graphs, such as DBpedia, Freebase, etc. These knowledge graphs are used in many artificial intelligence applications, such as question answering systems, semantic search, and relationship extraction. have important applications. As knowledge graphs become research hotspots, uncertainty is naturally introduced into the research of knowledge graphs: on the one hand, due to the fact that all aspects of graph construction, such as relationship extraction and entity linking, are full of noise and errors, it is difficult to Guarantee the certainty of the final knowledge map; on ...

AI Technical Summary

Problems solved by technology

Currently, representative methods that combine ontology and knowledge graph representation learning include JOIE, etc. However, the ontology in these works requires manual construction, which is time-consuming and laborious.

In order to solve this problem, some methods hope to use rule mining technology to automatically mine ontology from knowledge graph. However, these mined ontology are often of low quality and have great uncertainty.

Images