US-SQL: Managing Uncertain Schemata (demonstration proposal)

ABSTRACT In this paper we describe a demo concerning the manage-ment of uncertain schemata. Many works have studied theproblem of representing uncertainty on attribute values ortuples, like the fact that a value is 10 with probability .3or 20with probability .7, leading to the implementation ofprobabilistic database management systems. In our demowe deal with the representation of uncertainty about themeta-data, i.e., about the meaning of these values. Us-ing our system it is possible to create alternative proba-bilistic schemata on a database, execute queries over uncer-tain schemata and verify how this additional information isstored in an underlying relational database and how queriesare executed. Categories and Subject Descriptors H.2 [DATABASEMANAGEMENT]: Miscellaneous General Terms Management, Languages 1. INTRODUCTION The relational model makes a clear distinction betweentwo features of the data, separating their extension, or in-stance, i.e., the values taken by the attributes, from theirintension, i.e., the schema, or meta-data. This separation isalso reflected in the way in which data is stored and manip-ulated inside relational database management systems.Existing uncertain database management systems havebeen developed to represent and manipulate uncertain in-stances: attributes or tuples are annotated with probabil-ities and additional information to represent their proba-bilistic dependencies. Some relevant probabilistic databasemanagement systems are described in [3, 4, 10, 2] and fol-lowing works, and a collection of surveys on uncertain datamanagement can be found in [1].