语义OLAP缓存技术研究

Research of Semantic OLAP Caching

针对页面缓存,元组缓存,语义缓存等查询缓存技术不能充分利用OLAP较一般SQL查询所具有更强的语义和上下文相关性的特性,提出一种新的OLAP查询缓存技术语义OLAP缓存.该缓存模型只存储等价类所代表的集合信息,而不是大量单个数据单元或数据元组信息.随着查询过程的深入,缓存项的等价类范围动态更新,可在保持缓存大小不变的情况下,扩展等价类范围,从而回答更多查询并提高命中率.此外还提出根据缓存项间的上卷、下钻关系将缓存项连接在一起形成代数格结构,以方便剪枝,从而进一步缩小查找范围.实验结果充分证明了该缓存模型和算法的有效性.

The query caching techniques such as page caching, tuple caching and semantic caching can not take full advantage of theOLAP features like stronger correlation between the semantic relevance and contextual relevance. Thus in this paper we propose anew kind of caching technique semantic OLAP caching. This cache model keeps the succinct cache structure which stores equivalentclasses instead of amount of individual data cells or data tuples. With more queries submitted by users, the cache structure is adapteddynamically and the range of the equivalent class is extended, so more answers can be obtained when queries and the hit ratio can beimproved while the memory space is fixed. Further more, these cached items will join together to form a quotient cube lattice accordingto their roll-up/drill-down relationship, which reduces scanning time by pruning. The experiment proved the effectiveness of the se-mantic OLAP caching model and algorithm.