基于范围语义的非一致性数据库聚集查询  

Aggregation queries based on range semantics in inconsistent databases

作　　者：谢东[1] 吴敏[1] 

机构地区：[1]中南大学信息科学与工程学院,湖南长沙410083

出　　处：《中南大学学报（自然科学版）》2008年第4期810-815,共6页Journal of Central South University:Science and Technology

基　　金：湖南省自然科学基金资助项目(07JJ6113;07JJ3119);湖南省教育厅科研基金资助项目(07C832;07C385)

摘　　要：基于非一致性关系数据库的非聚集查询技术,发展普通的一致性查询应答的语义,提出非一致性数据库的范围语义,给出基于一致性查询范围的实际聚集查询重写方法,返回基于这种语义的聚集查询范围值。通过对初始分组属性和键属性聚集得到候选结果集,再过滤与初始分组属性值相冲突和不满足初始查询的元组。对于最小值,过滤掉可能不一致的元组,去掉可能值,得到一致性值;对于最大值,得到可能的最大候选值,实验基于TPC-H基准采用不同的参数进行研究。研究结果表明,聚集属性和投影属性数量、数据库以及初始查询的结果集对重写查询的负载有显著的影响,但非一致性数据比例和共享共同键值元组数的影响很小。尽管重写查询比初始查询的执行时间长,但是可以接受。Based on the query technique without aggregation in inconsistent databases, the common semantics of consistent query answering was developed to present the range semantics and a practical rewriting approach for aggregation queries based on the range of consistent answer, which returned to the range values. The candidate set was obtained by aggregating the original grouping and key attributes, and filtering the tuples that conflicted with the original grouping attribute values and were dissatisfied with the original query. For the minimum, the approach filtered possible inconsistent tuplcs and discarded the possible values to obtain the consistent values. For the maximum, the possible greatest candidate values were obtained. In the experiment, TPC-H was used to study the performance with different references. The results show that the overload of rewritten query is obviously affected by some aspects such as the number of aggregation and projection attributes, the size of database and result sets of the original query, but the overload has little effect on the proportion of inconsistent data and the number of tuples with the same key values. The running time of rewritten queries is longer than that of the original queries, but the overload is reasonable and the approach is effective.

关 键 词：关系数据库 非一致性数据库 聚集查询 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]