Concepts approximation for information retrieval from databases and pragmatic OLAP

This paper presents rough SQL, which is a prototype, based on an extension of the SQL algebra by introducing concepts approximation operators. This extension uses mainly the lower and the upper approximations of the Rough Set Theory (RST). The SQL algebra is a powerful tool to extract explicit information stored in databases. This extension is then used to support flexible querying of multidimensional OLAP data and rules discovery. Rough SQL is based on an efficient and optimized algorithm that operates approximations on data only once and stores the results in the database. The advantages are: the first is reducing considerably time spent to do the same treatment for each concept approximation request of the user; the second is that users are so given answers immediately. It proposes just adding concept approximation operators to the SQL algebra for more flexibility when dealing with imperfect data. Rough SQL can be connected to any relational database and applied on data warehouses. These capabilities allow extending the OLAP (on line analytical processing) analytical querying capabilities with approximate queries supported by the Rough SQL.