Abstract: Generalized view is a new form of data publishing. The difference between generalized views and classical materialized views is that data in generalized views are produced by generalizing data of basic database. It becomes a critical problem how to maintain the consistency between generalized views and the basic database, when an update operation is occurring on basic database. At first, for each update operation of insertion, deletion and modification, this paper analyzes the possible update responses occurring on generalized views. Then, this paper proposes the concept of multi-dimension bucket, and gives the construction method and related operations of multi-dimension bucket. Based on this, corresponding to various update operations of basic database, this paper presents the update algorithms based on multi-dimension bucket for generalized views, and solves the problems of updated maintenance consistency for generalized views.

Key words: privacy protect, generalized view, multi-dimension bucket, update, consistency

摘要： 泛化视图是一种新出现的数据发布形式。与经典的物化视图不同，泛化视图中的数据是通过对物化视图中数据进行泛化操作得到的。当基础数据库发生更新操作时，如何对泛化视图进行一致性维护是一个亟待解决的问题。分析了泛化视图在对基础数据库进行插入、删除、修改更新操作时可能出现的更新反应；提出了多维桶的概念，并给出了多维桶的构造方法及相关操作；在此基础上，对于基础数据库的不同更新操作，分别给出了基于多维桶的泛化视图更新一致性算法，解决了泛化视图的更新一致性维护问题。

关键词: 隐私保护, 泛化视图, 多维桶, 更新, 一致性