Research And Implementation Of Entity Alignment Method For Public Cultural

With the prosperity of public culture and the rapid development of knowledge graphs,various cultural organizations build public cultural knowledge graphs to abstractly model the relationship between public cultural resources.However,the existing public cultural knowledge graphs are small in scale and low in richness,and cannot be effectively To support public cultural services,it is necessary to use the entity alignment method to find resource entities with the same semantics between different public cultural knowledge graphs,so as to lay the foundation for the integration of large-scale public cultural knowledge graphs and provide support for public cultural services.Early entity alignment relies on ontology inference matching rules or attribute similarity for entity alignment,but different knowledge graphs have different ontology structures and attribute types,and the method is weak in applicability.After that,researchers use the knowledge graph embedding model to generate entity embeddings containing semantic information,and obtain entity alignment results by comparing the similarity between entity embeddings.This kind of entity alignment method can be used in the field of public culture.However,there are few relation types in the public cultural knowledge graph.When the existing entity alignment methods are used in the public cultural knowledge graph,the generated entity embeddings are insufficient for semantic representation,resulting in low entity alignment accuracy.Therefore,in order to make the semantic representation of entity embeddings sufficient,this paper studies from two aspects of knowledge graph embedding model and entity alignment method.The main research work includes the following three aspects:(1)In order to enrich the semantics of entity embedding,a knowledge graph embedding model based on relational semantic enhancement is proposed.Firstly,the reciprocal relationship is introduced,the corresponding reciprocal relationship is added to the relationship in the public cultural knowledge graph,the relationship type in the public cultural knowledge graph is expanded,and the implicit semantic information of the knowledge graph is mined.Secondly,the initial embedding vector of the knowledge graph is generated by the Rotat E model that efficiently models the reciprocal relationship;then the relationship information is explicitly used for the generation process of entity embedding,and a graph attention network based on relational semantic enhancement is proposed;finally,through joint training The Rotat E model and relational semantic-enhanced graph attention network model generate semantically informative entity embeddings.(2)In order to improve the accuracy of entity alignment and reduce the system overhead,an entity alignment method based on adaptive multi-hop neighbor aggregation is proposed.First,the candidate alignment entity set of the target entity is screened according to the alignment threshold interval,and then for the target entity and candidate alignment entity,cross-graph weights and intra-graph weights are set to aggregate multi-hop neighbor entity information that meets the selection rules to obtain neighbor information embedding,fully capturing entities The similarity between the important neighbor information between pairs,and finally the neighbor information embedding and entity embedding are connected for entity alignment calculation,and the entity alignment result is generated.(3)In order to verify the above theoretical research,this paper designs and implements a prototype system of public cultural knowledge graph entity alignment,and integrates it into the public cultural resource knowledge graph system platform.Entity alignment is performed on a real public cultural dataset to verify the feasibility of the entity alignment method proposed in this paper.In summary,in order to better integrate the public cultural knowledge graph and improve the quality of public cultural services,this paper conducts an in-depth study on the problem of entity alignment of knowledge graph.Firstly,a knowledge graph embedding model based on relational semantic enhancement is proposed.On this basis,an entity alignment method based on adaptive multi-hop neighbor aggregation is proposed.Finally,a prototype system of public cultural knowledge graph entity alignment is designed and implemented,and through a series of comparisons Experiments and system experiments verify the feasibility and effectiveness of the research results in this paper.