A technique for mutual inconsistencies detection and resolution in virtual data integration environment

Abstract

Data Integration refers to the problem of combining data residing at homogeneous, autonomous, and heterogeneous data sources, and providing users with a unified global schema. Users pose their queries in terms of this unified global schema. Data integration system allows users to perceive the entire collection as a single source, query it transparently, and receive a single and unambiguous answer. This entire collection of data sources may conflict with each other on three levels: schema level, data representation level, or data level. Most of the approaches in this area of research resolve conflicting among different schemas and among different data representations, and ignore the possibility of data-level conflict altogether; where most of the proposed techniques didn't deal with mutual inconsistencies over consistent data sources. Other techniques deal with mutual inconsistencies but with a set of drawbacks. In this paper a technique for detecting and resolving mutual inconsistencies between data sources is introduced where we assume that the schema level inconsistencies are resolved and the representation of data are unified over all data sources