This paper presents a system for representing temporal knowledge, expressing temporal queries, performing temporal reasoning chores, and carrying out temporal reason maintenance. The system can be regarded as an extension of conventional predicate calculus databases such as those used by Prolog [1] and reason maintenance systems such as the one described by Doyle [2].

Although its length (55 pages) may be daunting, this important and well-written paper should be read by everyone interested in computational approaches to time. The paper contains seven sections. Section 1 introduces the subject matter. Section 2 proposes “shallow temporal reasoning” as a way of dealing with combinatorial problems associated with temporal reasoning. Section 3 introduces the “time map,” a graphical representation for temporal information. Sections 4 and 5 form the bulk of the paper. Section 4 deals with time map management, including notation, basic temporal notions, default reasoning about time, deduction involving forward and backward chaining, noticing assumption failures, and automatic projection and refinement. Section 5 deals with temporal reason maintenance, including an algorithm for updating the temporal data dependency network and a proof of correctness for the algorithm. Section 6 mentions some problems with the approach proposed in the paper and offers suggestions for further work. Section 7 presents the conclusions.

The paper assumes that the reader has a general knowledge of deductive retrieval, logic programming issues such as unification, and Prolog. It will be of interest to graduate students and researchers in database management and artificial intelligence.