摘要
具有多维属性的实体相互连接构成的网络(如社交网络)称为多维网络,在多维网络上支持联机分析处理具有重要的应用价值。现有方法大都从文件或数据库中逐条读取记录,当数据量很大时,需要多次读取磁盘,导致查询响应时间过长,效率较低。文中提出了一种新的基于倒排索引的多维网络存储模型II-GC(Inverted Index based Graph Cube),通过将图的拓扑结构和顶点的多维属性存储在倒排索引列表中加快查询速度,并给出了在多维网络上进行聚集查询(cuboid)和交叉查询(crossboid)的算法。在DBLP数据集上的实验表明,该模型较Graph Cube的查询效率更高,扩展性更好。
A network such as social network linked by entities with multiple attributes is called multi- dimensional network. OLAP query on multi- dimensional network has an important application value. Most existing methods read records one by one from a file or a database. When a lot of data involved,these methods need more I / O time,thus leading to large query response time and low query efficiency.A newmulti- dimensional network storage model based on inverted index is presented,called II- GC（ Inverted Index based Graph Cube）. It speeds up the process by constructing inverted index both on topological graph and multiple attributes. Algorithms about cuboid and crossboid are also introduced. Experimental results on DBLP show that this model is more efficient and scalable than Graph Cube.
出处
《计算机技术与发展》
2016年第4期25-30,共6页
Computer Technology and Development
基金
国家自然科学基金资助项目(61201414
61301245
U1233113)
