WebView的物化选择和自适应更新

View-Selection and Self-Adaptive Update of Web View

1.引言 随着Internet的发展和异构信息源(从传统的RDBMS到半结构化信息源[13])集成技术的提高,对Web环境下的信息进行集成、分析处理并提供决策服务已成为当前的研究热点.Web信息集成最初采取Mediator/Wrapper框架的虚拟方案.Ullman在提交给VLDB‘97的论文[14]中指出,"Mediator实际上是一个或多个信息源上的视图".

Web Views mean Web pages that are stored in a Web repository. As it will bring users high efficiency .Web-View is very useful for some systems, especially those have many OLAP (OnLine Analytical Processing) applications. However,it's difficult to tell which pages should be stored and how to maintain them. When the update frequency of Web View is low.it can decrease query response time and reduce network overloading to materialize Web View. Unfortunately, this is not the real case in a fast changing Web environment. Given high update frequency.if every update to base data leads a WebView refresh.the system will lose its performance. On the other hand,if Web View does not change accordingly .the system will risk running out of date and becoming useless. Thus we are in a dilemma and must balance between them. In this paper we introduce a profit-based strategy and give its approximate algorithm. According to the experimental data,our algorithm is scaleable and adaptive to the Web environment and better than traditional policy such as virtual algorithm and pure materialized algorithm.