The automation of reasoning about the consequences of actions has been viewed as an important problem since the early days of Artificial Intelligence. Logic programming turned out to be an excellent tool for describing actions and their effects. The use of negation as failure leads to a simple solution to the frame problem. Unlike the first attempt to solve the frame problem using circumscription, the logic programming formulation does not fall into the trap known from the "Yale shooting" example. Moreover, some of the reasoning tasks related to actions can be solved by simply running the corresponding programs under Prolog, without developing any additional algorithms for nonmonotonic reasoning.
Research on representing action by logic programs has demonstrated the potential of several extensions of "classical" logic programming such as the use of abduction, disjunctive programs, and programs with two negation operators. In several ways, it has contributed to the development of the art of representing knowledge in logic programming.
The papers included in this special issue describe the results of some of the recent work on the connections of logic programming to the theory of action.
The contributions by Van Belleghem, Denecker, and De Schreye and by Kowlaski and Sadri analyze the relationship between two ontologies and notational systems for describing change--the situation calculus and the event calculus. They suggest ways to make both formalisms more expressive.
The paper by Levesque, Reiter, LespErance, Lin, and Scherl describes a logic programming language whose intended applications include the control of robots and industrial processes. The language is based on a theory of actions expressed in the situation calculus.
In the next group of papers, the authors develop a series of new high-level action languages and relate them to logic programming. Baral and Gelfond are interested in the concurrent execution of actions, and especially in the cases when