Knowledge Graph Generation Algorithm Based on Bayesian Network

The knowledge graph is a structured knowledge base used to describe concepts, entities, and their interrelationships in the physical world in symbolic form. Compared to large models, knowledge graphs have more explicit knowledge representation and structural advantages. Having stronger data integration and processing capabilities, reasoning and interpretability. This article is based on the Bayesian network structure learning algorithm MMPC (Max-Min Parents and Children), and constructs a knowledge graph through three modifications: Firstly, we use text segmentation and calculation of intersection to union ratio to preliminarily measure the coupling relationship between entities, reduce irrelevant entities, and lower computational complexity. Besides, mutual information clustering algorithm is used to improve the accuracy of independence judgment. Finally, we use a greedy algorithm based on information entropy gain ratio to find the optimal network structure.