Knowledge graph extrapolation method and system based on multilayer path perception

• Main Authors: JIN HAI, ZHAO FENG, CHEN MINGTAO, LIU KANGZHENG
• Format: Patent
• Language: Chinese; English
• Published: 07.03.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention relates to a knowledge graph extrapolation method and system based on multilayer path awareness. The knowledge graph extrapolation method comprises the following steps: learning embedded representation of entities, relationships and timestamps by applying a relational graph convolutional network encoder, and capturing dynamic evolution of facts; designing an emerging task processing unit to construct a multi-layer entity set, and matching corresponding historical relevancy for the entity sets of different layers; dividing the prediction task into a plurality of reasoning scenes, and accessing each reasoning scene to a processing unit of a corresponding level to complete division of a multi-layer historical related entity set; a multi-classification task solving method is used for obtaining prediction probability distribution of a target entity, the entity with the highest probability serves as a prediction answer to complete an extrapolation task of the time sequence knowledge graph, and the pre