基于列数据库和图缓存的海量RDF管理

Massive RDF Management Based on Column Oriented Database and Graph Cache

针对现有的基于关系数据库和图数据库存储RDF数据集的几种模型中查询性能的不足,将列数据库和图缓存相结合,提出一种新的管理海量RDF数据的方案.该方案在底层磁盘采用基于列的关系存储,同时在内存中构建RDF图模式的存储,并设计实现了一套新的SPARQL查询引擎.通过相关分析和各种存储模式的实验结果对比分析,验证了该方案的可行性,表明了该方案具有更高的查询效率.

Aiming at the poor query performance of several models based on relational databases and graph database for RDF dataset storage,a new method is given to store and manage the massive data efficiently by combining the column-oriented model and graph model. RDF data is stored in disk based on column-oriented relational model. Part of RDF data are loaded in memory as graph schema. Besides,a SPARQL query engine is designed and implemented. According to results of the comparative experiment,this method is practicable and more efficient in query performance.