Knowledge Representation Learning Method Integrating Relationship Path, Type and Entity Description Information

Abstract

The invention discloses a knowledge representation learning method fusing relationship path, type and entity description information. The method comprises the following steps: (1) preprocessing a knowledge base to extract structured information, relationship path information, type information and entity description information; (2) selecting a positive sample from the knowledge base, constructinga negative sample, calculating the loss of the structured information, the relationship path information, the type information and the entity description information corresponding to the positive andnegative samples; (3) updating parameters, a practice vector and a relationship vector of a model according to the loss; and (4) repeating the step (1) to the step (3) until the number of iterations reaches the preset maximum number of iterations, and outputting the vectors of the entity and relationship to realize the learning of the knowledge representation. According to the method, on the basisof using a relationship path, modelling is performed on the knowledge base by fusing multi-source information, so that the problem that the only the structured information is used in traditional knowledge representation learning method and multiple additional information is not adopted is solved.

Description

technical field [0001] The invention relates to the field of knowledge representation learning, in particular to a multi-source information fusion knowledge representation learning method. Background technique [0002] With the advent of the Web3.0 era, the knowledge base containing a large amount of structured knowledge has become an important part of many semantic applications (such as intelligent question answering and search services), and more and more enterprises and organizations are committed to constructing large knowledge bases. The structured knowledge in the knowledge base is represented in the form of triples (head entity, relation, tail entity). Although the existing knowledge base already contains a large number of triples, there are still missing entities and relations in the knowledge base because the information is massive and constantly changing. How to complete the missing relationship between entities in the knowledge base is a key issue in the construc...

AI Technical Summary

Problems solved by technology

For example, when using relational path information, the discrimination of relational paths is poor, which blurs the semantics of relational paths; when using type information, only the types of entities and relations are considered, but the types of relational paths are not considered; in the existing model At most, only one kind of additional information is used, and multi-source information is not used, resulting in a waste of information

Images