Proceedings of the 12th International Joint Conference on Artificial Intelligence (IJCAI'91), Sydney, Australia, 24-30 August 1991, pages 1177-1182. ISBN 1-55860-160-0.
Abstract
Reification of propositions expressing states, events, and properties has been widely advocated as a means of handling temporal reasoning in AI. The author proposes that such reification is both philosophically suspect and technically unnecessary. The reified theories of Allen and Shoham are examined and it is shown how they can be unreified. The resulting loss of expressive power can be rectified by adopting Davidson's theory in which event tokens, rather than event types, are reified. This procedure is illustrated by means of Kowalski and Sergot's Event Calculus, the additional type-reification of the latter system being excised by means of a general procedure proposed by the author for converting type-reification into token-reification. Some examples are given to demonstrate the expressive power of the resulting theory.
Read the paper (Postscript file, 11 pages)