Research on Semantic Similarity of Entities with the Case of Event Knowledge Graph

Evaluating entity similarity is a common problem in recommendation systems. In the business field, entity similarity generally uses a collaborative recommendation calculation method to recommend object features and user features. However, in the military field, especially for the recommendation of military knowledge graph entities, more consideration needs to be given to the information implicit in the association between entities. Therefore, the existing similarity calculation model needs to be supplemented and improved. Knowledge graphs have important practical significance in the management, visualization, analysis and reasoning of military data. On the basis of the knowledge graph, we propose a framework for calculating the asymmetric similarity of the domain knowledge graph, using as much as possible the attribute constraints of the ontology design and the convenience of long-distance reasoning of knowledge graph. War consists of various campaign events and this article takes the war knowledge graph as an experiment example. This model WKGSM that has been proposed draws on the traditional entity similarity calculation framework, considering the attribute value of the entity, the relationship between the entities, and the number of sub-entities. The framework adopts a three-layer structure to decompose big problems into small ones. Finally, the three-layer scores are merged through the AHP. The parameters of the fusion function and the coefficients of the similarity matrix can be modified according to the needs of the actual problem. The experimental results show the effectiveness of the method. In addition, this article also discusses the use environment and limitations of the model.