A Probabilistic Relational Data Model for Uncertain Information

This paper introduces a probabilistic relational data base model (PRDB). In this model, the tuple attribute value of relations is uncertain and represented by a probabilistic triple. Every probabilistic triple consists of a set of relational attribute values and two upper and lower bound probability distributions over the set of the attribute values. Semantics of such representation indicating that the attribute has a value in the set with a probability belongs to an interval from the upper and lower bound of the set. On the base and using probabilistic combination strategies in Eiter et al (2001), we define the notions of uncertain values, schemas, probabilistic relations, and selection operation as respective extensions of the notions of values, schemas, relations, and selection operation in the conventional relational database model.