支持复杂语义的数据流滑动窗口连接建模和查询优化  

Complex Semantically Supported Modeling/Optimization of Joining Execution of Sliding Window Queries over Data Streams

作　　者：谷峪[1] 李晓静[1] 许嘉[1] 于戈[1] 

机构地区：[1]东北大学信息科学与工程学院,辽宁沈阳110004

出　　处：《东北大学学报（自然科学版）》2008年第11期1552-1555,共4页Journal of Northeastern University(Natural Science)

基　　金：国家自然科学基金资助项目(60773220;60773222)

摘　　要：数据流连接操作的执行方式直接影响到查询执行的效率和结果的正确性.在STREAM项目流化方法的基础上,对复杂应用下的数据流滑动窗口连接的准确语义进行了分析和建模,以时间窗口连接的增量维护为例提出了两种改进的基本维护策略,给出了针对元组窗和分组窗的维护方法,讨论了根据参数和环境的不同进行适应性的维护.实验证明了在不同的窗口、流速和选择度下,适应性地选择不同的优化算法会达到最优的处理性能.How the data stream is executed continuously will affect the efficiency of the query and correctness of its result. Based on the relevant streaming methods of the STREAM project, the semantic exactness during the joining execution of sliding window queries over data streams is analyzed and modeled under complex conditions, and two improved basic maintenance strategies are proposed taking time-based window as example. Furthermore, the maintenance schema for tuple-based and partition-based windows is illustrated and the adaptive maintenance in accordance to different parameters and environments is discussed. Experimental results showed that under conditions of different window, stream rates and selectivities the optimization algorithm to which the adaptivities are differently chosen will have the ideal processing performance.

关 键 词：数据流 复杂语义 滑动窗口 连接 查询优化 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]