摘要
Data cube computation is a well-known expensive operation and has been studied extensively. It is often not feasible to compute a complete data cube due to the huge storage requirement. Recently proposed quotient cube addressed this fundamental issue through a partitioning method that groups cube cells into equivalent partitions. The effectiveness and efficiency of the quotient cube for cube compression and computation have been proved. However, as changes are made to the data sources, to maintain such a quotient cube is non-trivial since the equivalent classes in it must be split or merged. In this paper, incremental algorithms are designed to update existing quotient cube efficiently based on Galois lattice. Performance study shows that these algorithms are efficient and scalable for large databases.
Data cube computation is a well-known expensive operation and has been studied extensively. It is often not feasible to compute a complete data cube due to the huge storage requirement. Recently proposed quotient cube addressed this fundamental issue through a partitioning method that groups cube cells into equivalent partitions. The effectiveness and efficiency of the quotient cube for cube compression and computation have been proved. However, as changes are made to the data sources, to maintain such a quotient cube is non-trivial since the equivalent classes in it must be split or merged. In this paper, incremental algorithms are designed to update existing quotient cube efficiently based on Galois lattice. Performance study shows that these algorithms are efficient and scalable for large databases.
基金
国家自然科学基金，国家高技术研究发展计划(863计划)，国家重点基础研究发展计划(973计划)，教育部重点科研项目
