Ongoing Research Seminar
Friday April 26, 1996

Michael Kifer
Deductive Object-Oriented Databases:
From Wishful Thinking to Virtual Reality

According to rumors, the early hybrids of object-oriented and deductive languages were mutants that escaped from secret Government AI labs. Whether this is true or not, the fact is that by mid-80's, database and logic programming communities began to show interest. The temptation was hard to resist: the object-oriented paradigm provides a better way of manipulating structured objects, while logic and deduction offer the power and flexibility of ad hoc querying and reasoning. Thus, hybrid languages have the potential for becoming an ideal turf for cultivating the next generation of information systems.

There are two distinct aspects in object-orientation: structural (the statics) and procedural (the dynamics). The static aspect was somewhat easier to capture in logic and after taking a few wrong turns we developed F-logic. Taming the dynamics was harder, because there was no widely accepted solution to the problem of state changes even for traditional deductive languages. Our recent work on Transaction Logic appears to provide an adequate framework for specifying the dynamics in classical deductive languages.

In this talk, I will first review some core aspects of F-logic and Transaction Logic, and then show how the two can be combined in a natural way to yield a unified foundation for future work on deductive object-oriented languages, both in theory and practice.