基于标记的不一致数据查询处理框架

为能在函数依赖范围内准确定位查询结果中的不一致数据,采用查询重写的方法建立基于标记的不一致数据查询处理框架.该框架是RDBMS与用户之间的一类中间件,以标记区分源数据和查询结果中的可信和不可信部分,可以避免信息丢失;能将任意传统SQL查询翻译成能返回带信任标记的查询结果的SQL查询集,由已有的RDBMS响应.该系统能够内嵌到现有数据库应用系统中,且用户无须掌握新查询语言.实验表明,该系统不仅在信息保存上与相关工作相比有很大优势,而且时间性能与传统查询相比相差无几.

A framework of annotation based query process over inconsistent data

To mark out inconsistent data from query results acurrately in the scope of functional dependency, a framework named annotation based query answer over inconsistent data is developed using the methods of query rewriting. The framework is a middleware between RDBMS and users, which distinguishes incredible data from credible ones in data source and query results with annotations, so as to avoid information loss. The framework can translate any traditional SQL queries into a set of SQL queries which can return query answers with annotations, and the query is answered by existing RDBMS. The framework can be easily embedded into the existing database application system, and users need not have to master a new query language. The experimental results show that the framework has not only overwhelming advantage against related works in information preservation, but also almost the same time performance with traditional queries.