Research And Implementation Of Semantic Query And Inference Based On Graph

As an important data structure in discrete mathematics and computer science, graph models have begun to be used by more and more emerging technology fields to represent complex data entities in the real world and the relationships between those entities. At the same time, Complex networks with uncertain can also be modeled as probabilistic graphs.However, the rich semantic features and complicated internal structure of the graph data have greatly increased the complexity of graph data manipulation algorithms. With the increase of the data volume, it is significant to improve the performance of the graph data manipulation algorithms, especially for the large graph data. Therefore, this paper makes a research on semantic query and reliability reasoning algorithm based on graph.Based on the semantic and structural features of real-life graphs such as social networks, this paper presents a novel approximation sub-graph matching query algorithm named ASMQ on large graph, which can relax the rigid label and topology matching constraints of sub-graph isomorphism. In ASMQ, the sub-graph matching measurement model is firstly introduced, and the matching degree between query graph and matched sub-graph is measured from two aspects: graph structure and node labels. Then the candidate nodes are filtered from the target graph according to the Neighborhood Domain Index. Finally, a heuristic inference method is used to determine the best matching sub-graph from the candidate set. Experiments show that our ASMQ algorithm outperform state-of-the-art approximate graph query algorithm NeMa on query precision and query efficiency.In uncertain graph, this paper designs a relational reliability reasoning algorithm, and based on the algorithm, we can get all the nodes in uncertain graph, whose reliability from the given source node set is greater than the given probability threshold and semantic similarity threshold. In the algorithm, a hierarchical clustering of uncertain graph is firstly carried out by using METIS to construct SRR-Tree index. Then the SRR-Tree index is used to filter candidate nodes according to the reasoning conditions and semantic information of node labels. And finally, two verification methods are used to determine the final inference result from the set of candidate nodes. The two verification methods are respectively named lower bound based verification and sampling-based verification. Experiments show that our algorithm has good reliability reasoning performance.In this paper, based on our graph query algorithm, a visualized sub-graph matching query system is developed to display the query results in a more intuitive way, so that more useful information can be found on the graph data.