【摘要】 在关系数据库系统中，视图是从一个或几个表中导出的表，为了提高访问视图数据的效率，人们将视图数据也保存起来，这样的视图称为实视图．当基表发生了变化，为使实视图与基表保持一致，必须对实视图进行维护．实视图的维护可分为三个阶段：即无关更新检测、可自处理更新检测和实际维护．人们企图通过前两个阶段检测出不需要维护的更新和不访问基表数据即可进行维护的更新，这样可以有效地提高实视图维护的效率．本文针对复杂实视图，采取利用语法树进行运算分解的策略提出了检测这两类更新的解决方案，并给出了相关的算法更多还原

【Abstract】 In relational database systems,a view is a relation derived from one or more relation.To access views efficiently,we can store the data of them.Such views are called materialized views.When the base relation changes,a materialized view should be maintained so that it becomes consistent with the base relations.Generally,the maintenance of a materialized view can be split into three steps,i.e.,detection of irre levant updates,autonomously computable updates and maintenance operations.The first two steps are used to find updates that cause no maintenance operations on materialized views,or cause maintenance operations which need not access the base relations.In this paper,we provide solution to detection of these updates on complex materialized views based on operator decomposition,and also provide corresponding algorithms.更多还原