A uniform methodology for extracting type conflicts and subscheme similarities from heterogeneous databases

摘要

Cooperative Information Systems have been proposed to allow a uniform access to heterogeneous data yet preserving their operational autonomy. They use global dictionaries defined on the basis of interscheme properties; these include nominal and structural properties, type conflicts and object cluster similarities. Whereas in the literature a certain number of techniques has been proposed for deriving nominal and structural properties, few approaches exist for detecting type conflicts and object cluster similarities. The type of an object indicates if it is an entity, a relationship or an attribute; type conflicts indicate the existence of objects representing the same concept yet having different types. Object cluster similarities denote similitudes between portions of different schemes. This paper proposes an automatic, probabilistic approach to the detection of type conflicts and object cluster similarities in database schemes. The method we are proposing here is based on considering pairs of objects having different types (resp., pairs of clusters), belonging to different schemes and on measuring their similarity. To this purpose object (resp., cluster) structures as well as object (resp., cluster) neighborhoods are analyzed to verify similitudes and differences. A number of examples shows the suitability of our techniques to effectively detect type conflicts and object cluster similarities.