中文摘要：

数据服务屏蔽了数据访问的复杂性,并且支持用户以可视化方式定义装载异构数据源的嵌套视图.然而,当异构数据源输出的数据发生更新时,该更新需要通过数据服务逐层传播到顶层的嵌套视图,一类挑战性问题是,如何减少更新传播过程中的冗余计算,提高嵌套视图的数据新鲜度.为此文中提出了基于数据服务的嵌套视图动态更新方法,该方法利用指针为嵌套视图中的元组建立嵌套任意层次的数据服务的引用,同时给出了一种记录数据服务更新的日志以及在该日志上的嵌套视图增量更新算法.文中在原型系统中实现了该算法并且进行了实验分析,实验结果表明,该方法减少了70%的嵌套视图的更新时间,提高了60%的嵌套视图的数据新鲜度.

英文摘要：

Data service hides the complexity of the multi-source and heterogeneous data sources, and helps the implementation of user-steering nested view generation. However, once heteroge- neous data sources are updated, the updates must be propagated to the top-level nested views through data services step by step. A challenging problem is how to reduce the redundant compu- tation and improve the data freshness. In this paper, a dynamic update strategy is proposed. The strategy uses the pointer to establish the relationship between the data services and the nested views, and designs the log data structure for data service, and proposes an incremental update algorithm based on the log data structure. Experiments show that our strategy can effectively reduce 70% redundant computation, and improve 60% data freshness.