基于存储优化模型的RDF数据查询机制研究-计算机应用技术专业论文.docxVIP

华 华 中 科 技 大 学 硕 士 学 位 论 文 II II Abstract Traditional RDF (Resource Description Framework) storage systems use relational database system to manage RDF data using a triple table, but due to the sparse characteristic of RDF data, those storage systems did not show efficient storage usage and high query performance. Thus, some storage optimization methods are proposed, but they do not use schema information to optimize the storage model. SPARQL (Simple Protocol and Query Language) is a World Wide Web Consortium recommended RDF query language, how to efficiently support SPARQL queries in a particular storage model becomes a research hotspot. DBLinkQL is a memory based RDF data storage system that designed for efficiently SPARQL query. In order to make storage efficiency and high query performance, firstly DBLinkQL builds the string and integer identifier mapping module that can transfer strings into integer identifiers and vice versa; secondly DBLinkQL extracts special vocabularies from RDF schema space file, classifies the vocabularies into different sets and utilizes the vertical partition approach to do multiple optimization that improving the query performance. DBLinkQL supports a variety of specifications described RDF data imported into storage model which can be mapped into disk image, from which the memory storage model can be re-construct quickly. DBLinkQL has partially supported SPARQL by converting SPARQL query into query graph model that can be operated by the underlying storage model and handling different query situations; DBLinkQL builds the join cost model by getting the statistics of the underlying storage data and uses greedy algorithm to implement a query optimization module which can reorder the sequence of join operations to enhance the query performance. The experiment of testing the SPARQL query optimization module shows that the module greatly improved the performance of query. The experiments of performance show that DBLinkQL has higher perform