Studies On Fuzzy RDF Data Matching Based On Graph Model

With the rapid development of Internet applications and the deep research of Semantic Web technology,more and more application systems choose RDF(Resource Description Framework)as the medium for data representation and exchange.As the data foundation for semantic representation and reasoning on the Semantic Web,RDF data model is very generic and can describe and organize all resources on the Semantic Web in the form of directed label graphs.RDF plays an increasingly important role in the current Semantic Web environment and it has actually become the cornerstone of intelligent information processing in the Semantic Web era.In this context,the problem of effective RDF data management has attracted the attention of academia and industry,and theories and techniques of the extension of RDF data model and queries have become research hot topics in the database field.There are a lot of inaccurate and uncertain information in real world applications.It has become a hot topic of current research to incorporate fuzzy information into data modeling and processing to meet real-world needs.Many Web applications need to process a large amount of fuzzy information,and existing RDF models cannot be directly used for fuzzy information representation and query.Therefore,the need for fuzzy RDF data management has arisen.Unfortunately,although some researchers have devoted to investigate fuzzy RDF data in recent years,the research on modeling and matching query of fuzzy RDF data model is still in its infancy stage.Many important theoretical issues and technical difficulties are still a research gap.In view of this situation,we propose an effective and efficient solution and related techniques along with fuzzy RDF graph data modeling in this thesis,storage and matching for constructing a robust system for fuzzy RDF graph data management.The specific research content of this thesis is as follows:(1)Aiming at the problem that the existing fuzzy RDF data model can only represent the tuple-level fuzzy information,a fuzzy RDF graph data model capable of supporting element-level fuzzy information is proposed,and the formal definition of the model is given.Based on the model,a general fuzzy RDF graph algebra framework is proposed.This algebra framework can effectively support inaccurate and uncertain RDF information query.Furthermore,we introduce some algebraic equivalence for optimizing queries in fuzzy RDF.With these as the basis,we introduce how to capture queries expressed in popular SPARQL query languages by using fuzzy RDF graph algebra expression.Fuzzy RDF graph data model and fuzzy algebra operations lay the foundation for the Semantic Web publishing of fuzzy data and the realization of intelligent information processing based on the Semantic Web.(2)For the query matching problem of fuzzy RDF graph data,a fuzzy RDF data matching method based on subgraph schema is proposed.First,based on the overall matching strategy,the subgraph pattern defined on the traditional RDF data model is extended to the fuzzy RDF subgraph pattern,that is,the fuzzy condition is specified on the vertex to indicate the user's query intent;the regular expression is marked on the edge,which represents the structure of the attribute path.Then a fuzzy RDF subgraph pattern matching algorithm based on extended graph homomorphism is proposed to evaluate the subgraph pattern on this fuzzy RDF graph.The method is based on a backtracking search method that generates each possible mapping by traversing the parse tree in a depth-first order.The implementation of the regular expression-based attribute path query method not only maintains the simplicity of the attribute path expression,but also realizes the query intent of identifying the user and returns the query result with the user preference.(3)Aiming at the problem that the subgraph pattern query algorithm may face too high cost,a path-based fuzzy RDF approximation subgraph query method is proposed in this thesis.It is inspired by the method of path instead of vertex as the basic matching object.In the preprocessing stage,the path index of the data graph is established,and the related information of the data graph path and its membership degree is saved.On this basis,an approximate fuzzy RDF subgraph matching algorithm is proposed.First,the algorithm decomposes the query graph into a set of possible overlapping paths,and then finds the matching paths in the data graph,and selects the path that best matches the query path as the candidate path subset.Next,the candidate path is joined together to build the answers of the query graph.Furthermore,the absolute membership of matching the result graph is calculated by aggregating the relative membership of each candidate path in the matching process.Finally,a comprehensive experiment demonstrates the effectiveness of the proposed method.(4)Aiming at the problem of alignment of two RDF graphs in a fusion and interconnection environment,a fuzzy RDF graph matching method based on similarity is proposed.In the method,we not only make use of the label information of vertices and edges,but also consider the structural features of the RDF graph.First,based on the label content of the graph elements(vertices and edges),the syntax and semantic similarity of the RDF graph vertex and edge labels are calculated respectively.Then,based on the idea of similarity propagation,the adjacency relationship of the vertices is taken as the structural feature information,and the structural similarity is calculated iteratively by combining the fuzzy membership degree of the vertex and the edge.Next,individual similarities are aggregated together to obtain overall similarity.Finally,according to the predetermined threshold,the optimal alignment could potentially be obtained.The experimental results show that the method can be used to effectively measure the similarity of fuzzy RDF graphs in the various scenarios.