Query Answering in Relational Inductive Databases

Inductive databases can be viewed as a natural extension of traditional databases to contain not only persistent data but also the generalization of stored data, which are called patterns. The idea of inductive databases has been proposed originally as a support system for the knowledge discovery or data mining process. Many SQL-like languages have been designed and implemented to include mining operators in the SQL primitives. We percept the concept of inductive databases in a different angle. In stead of designing yet another inductive database system, we are looking for the deployment of an existing inductive query language and environment to support the database tasks. We focus on the task of query answering which has a high potential of being a beneficiary of the stored patterns in inductive databases. Our experimental results of query rewriting technique using induced patterns as a semantic knowledge confirm this advantage.