Building method and system used for knowledge obtaining model in knowledge graph

Abstract

The invention provides a building method used for a knowledge obtaining model in a knowledge graph. The method comprises the steps of constructing a first training set consisting of multiple text sentences as input data and a relationship between any two entities in each sentence in the knowledge graph, as a classification result, and training a first neural network; constructing a second trainingset consisting of triples in multiple knowledge graphs, and training a second neural network; by taking input data vectors obtained in the second neural network as attention features of the first neural network, building a relationship extraction model; by taking input data vectors obtained in the first neural network as attention features of the second neural network, building a knowledge representation model; and fusing the relationship extraction model and the knowledge representation model to obtain the knowledge obtaining model in the knowledge graph. According to the method provided bythe invention, the two task models of knowledge representation and relationship extraction are integrated at the same time, and the features of the knowledge graph and free texts can be comprehensively extracted, so that the model stability and accuracy are improved.

Description

technical field [0001] The present invention relates to the field of information processing, and more specifically, to a method and system for constructing a knowledge acquisition model in a knowledge graph. Background technique [0002] Knowledge graph, also known as knowledge base in some scenarios, is a knowledge system formed after structuring human knowledge in the real world. In a knowledge graph, a large amount of knowledge, such as information in open databases and encyclopedias, is usually expressed in the form of relational data sets. In a relational data set, basic facts are abstracted as entities, and related information such as rules, logic, and reasoning are abstracted as relationships between entities. If entities correspond to points and relationships correspond to edges, then these knowledge can be further presented in the form of graphs, so that they can be efficiently used by computers, and this is the significance of studying knowledge graphs. This mode...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the problem of low knowledge acquisition performance in the existing knowledge graph knowledge acquisition model, which only considers the corresponding relationship between text and graph, we provide a method and system for building a knowledge acquisition model in knowledge graph

Images