HANDLING DISCOVERED STRUCTURE IN DATABASE-SYSTEMS

Most database systems research assumes that the database schema is det ermined by a database administrator. With the recent increase in inter est in knowledge discovery from databases and the predicted increase i n the volume of data expected to be stored it is appropriate to reexam ine this assumption and investigate how derived or induced, rather tha n database administrator supplied, structure can be accommodated and u sed by database systems. This paper investigates some of the character istics of inductive learning and knowledge discovery as they pertain t o database systems and the constraints that would be imposed on approp riate inductive learning algorithms is discussed. A formal method of d efining induced dependencies (both static and temporal) is proposed as the inductive analogue to functional dependencies. The Boswell databa se system exemplifying some of these characteristics is also briefly d iscussed.