Editorial: Temporal representation and reasoning

Time is one of the most relevant topics in AI. It plays a major role in several of AI research areas, ranging from logical foundations to applications of knowledgebased systems. This special issue originates from a selected set of the papers presented at the TIME-96 International Workshop on Temporal Representation and Reasoning. TIME-96 was the third in a series of workshops set up with the goal of exploring key issues and major trends in temporal representation and reasoning. As TIME-96 program co-chairs, we received a total of 45 submissions, of which 20 were eventually selected by the program committee for full presentation at the workshop. These 20 papers have been further screened to identify those which proposed the most mature and significant research. As a result of this screening, 13 papers were selected and their authors were invited to submit extended and revised versions for possible publication in this special issue. Each of these papers has been carefully reviewed by three experts in the field. As a result of this further selection process, 7 papers were accepted and are included in this issue. The papers appearing in this issue span a wide set of subjects, covering many of the current topics of AI research about time. They range from traditional themes, such as temporal constraints and reasoning about actions and events, to emerging trends, such as time granularity and time in medical reasoning.1 The paper by Bacchus and Kabanza shows the usefulness of temporal logic for planning. While traditional planning concentrated on primitive actions and instantaneous states, the introduction of a temporal dimension allows one to reason about sequences of actions and/or about the sequences of states they generate. The temporal logics adopted by the authors are quite simple: their syntax contains the usual temporal modalities and their models are linear sequences of states. Nevertheless, they can be successfully used to specify search control information as well as to specify goals that are temporally extended over time, that is, to constrain the way the agent achieves its final goal. Furthermore, the authors show how temporal logic can be used to define history-dependent rewards to be assigned to states in order to generate optimal plans. Bettini, Wang and Jajodia present a general framework for describing time granularity. They identify the main parameters according to which the various models of