Knowledge base completion method based on PRMATC algorithm

Abstract

The invention relates to a knowledge base completion method based on a PRMATC algorithm. The knowledge base completion method comprises the following steps: S1, importing all fact triples and entitiesin a large-scale semantic network knowledge base KB and storing same in a distributed cluster Neo4j graph database; S2, constructing a BILSTM-CRF model, and training the BILSTM-CRF model; S3, recognizing entities on the two sides of the relationship and classifying through a trained BILSTM-CRF model, and obtaining a definition domain and a value domain of the relationship through conversion; S4,improving an FP-Growth algorithm; S5, mining a strong association rule implied between transactions; S6, converting the obtained definition domain and strong association rule of the relationship intoa Horn logic rule; and S7, obtaining new knowledge according to the obtained Horn logic rule, and adding the new knowledge into a knowledge base KB. According to the method of the invention, the representative knowledge base Horn rule can be efficiently found, meanwhile, the quantity and accuracy of the mining rules are better than those of other rule mining systems, and the knowledge base can bebetter complemented.

Description

technical field [0001] The present invention relates to the field of massive data storage and reasoning under the knowledge graph, and specifically relates to a knowledge base completion method based on the PRMATC algorithm. Background technique [0002] Mining Horn rules from the large-scale semantic network knowledge base, and then using these rules to help infer and add the missing knowledge in the knowledge base is one of the most effective means to realize the dynamic growth of the knowledge base. Association rule mining algorithm is one of the important algorithms in the field of data mining, and its purpose is to mine the implicit relationship between transactions. Traditional algorithms include April i or i algorithm [1] and FP-Growth algorithm [2]. Traditional association rule mining algorithms have achieved good results in small-scale data sets. However, with the rapid development of Internet technology in recent years, network data has shown explosive growth. Tra...

AI Technical Summary

Problems solved by technology

Traditional association rule mining algorithms have achieved good results in small-scale data sets. However, with the rapid development of Internet technology in recent years, network data has shown explosive growth. Traditional association rule mining algorithms cannot be calculated by a single node and run out of memory. and other problems, so that it cannot meet the needs of network big data

Images