Research On Semantic Query Algorithm In Knowledge Graph

With the explosive growth of data,the content of knowledge graph expands rapidly,which leads to the increasing scale of knowledge graph and the difficulty of query.Most of the existing knowledge graph query algorithms are isomorphic queries based on entity tags.Because of the one-sided entity tags,they can not accurately reflect the semantic relationship of knowledge graphs,which leads to the low semantic relevance of query results.At the same time,the efficiency of the existing query algorithm is low because of the large scale of knowledge graph.In view of the above problems,this paper improves the knowledge graph compression and query algorithm,and the specific work is as follows:Firstly,a graph compression algorithm based on the complete equivalence class is proposed.For the large-scale knowledge graph,the node is divided by comparing the information of the node ontology and the neighbor node to determine whether the node belongs to the same completely equivalent class,and then judge whether the multiple edges meet the two-way relationship of the edges and compress the edges.By the graph compression algorithm,the knowledge graph is compressed into a small knowledge graph,which is more convenient for query and storage.Secondly,the knowledge graph query algorithm based on node similarity is proposed,and the strategy of "location filter query" is used to query.In the positioning stage,the central node is selected according to the global and local influence of the query graph,and the query sub region is located according to the candidate set of this node,so as to further reduce the query scale of knowledge graph.In the filtering stage,the semantic similarity and structure similarity of the nodes are considered to filter the candidate sets,and the semantic correlation of the candidate sets is improved.In the verification stage,the nodes in candidate set are verified by the edge label isomorphism,and the result set is generated.The result set is sorted according to the semantic relevance of the query results,and the first k results with high semantic correlation are output.Finally,experiments are carried out on the Yago and DBpedia data sets,and the comparison and verification with the existing algorithms are carried out.