Showing papers by "Hector J. Levesque published in 2003"

Rational interaction as the basis for communication

Abstract: : This paper derives the basis of a theory of communication from a formal theory of rational interaction. The major result is a demonstration that illocutionary acts need neither be primitive, nor explicitly recognized. As a test case, we derive Searle's conditions on requesting from principles of rationality coupled with a theory of imperatives. The theory rests on a formal account of intention and distinguishes insincere or nonserious imperatives from true requests. A theory of purposeful communication thus emerges as a consequence of principles of action and interaction.

Persistence, Intention, and Commitment

Abstract: This chapter contains sections titled: Introduction, Methodology, Elements of a Formal Theory of Rational Action, Persistent Goals, Intention as a Kind of Persistent Goal, An End of Fanaticism, Conclusion, Notes, References

A tractability result for reasoning with incomplete first-order knowledge bases

Abstract: In previous work, Levesque proposed an extension to classical databases that would allow for a certain form of incomplete first-order knowledge. Since this extension was sufficient to make full logical deduction undecidable, he also proposed an alternative reasoning scheme with desirable logical properties. He also claimed (without proof) that this reasoning could be implemented efficiently using database techniques such as projections and joins. In this paper, we substantiate this claim and show how to adapt a bottom-up database query evaluation algorithm for this purpose, thus obtaining a tractability result comparable to those that exist for databases.

Logical foundations of active databases

Abstract: Classical database management systems (DBMSs) have been enhanced over the past fifteen years with the addition of rule-based programming to obtain active DBMSs. Active behavior is mainly characterized by a rule language and an execution model. Execution models go hand in hand with advanced transaction models (ATMs) which relax the so-called ACID (Atomicity-Consistency-Isolation-Durability) properties. Both rule languages and execution models have been proposed in an ad hoc way to deal with applications which are not easily implementable using the classical DBMSs. Therefore an open problem in this area is to formally account for active behavior using a uniform formalism. This thesis gives logical foundations to active databases using the situation calculus, a logic for reasoning about actions. Our approach appeals to theories in which one may refer to all past database states. We give a logical semantics to an ATM by specifying this as a theory of the situation calculus called basic relational theory, which is a set of sentences suitable for referring to past database states in the context of database transactions. We express the properties of the ATM as formulas of the same calculus. Such properties are logically entailed by the basic relational theory that captures the ATM. We illustrate our framework by formalizing various transaction models. Next, we introduce active relational theories , which extend basic relational theories by capturing typical aspects of active behavior found in existing active DBMSs. We give a formal semantics to various features of active behavior by using active relational theories. We capture the most popular active rules, the Event-Condition-Action (ECA) rules, as programs written in a situation calculus based logic programming language that accounts for parallelism and we also write transaction programs transaction programs that have active relational theories as background axioms. We also classify various execution semantics for ECA rules. Finally, we give a method for implementing active relational theories in Prolog.