Catalogue of Artificial Intelligence Techniques


Jump to: Top | Entry | References | Comments

View Maths as: Images | MathML

Temporal Logic

Keywords: Event Calculus, modal temporal logic

Categories: Inference and Reasoning , Knowledge Representation , Planning

Author(s): Han Reichgelt

Temporal logic deals with reasoning about time. There are three different approaches to temporal logic. The first approach is the Situation Calculus. In the situation calculus, one simply adds an argument to each predicate which represents the time at which the predicate is assumed to be true. Thus, a two-place predicate like `hit' becomes a three-place predicate, and the proposition that `Harry loves Mary at time t' is represented simply as hit(harry,mary,t). The second approach is the reified approach. In the reified approach, of which the Event Calculus is an example, one complicates the language by introducing separate terms for events, processes etc. as well as a predicate which is sometimes called HOLDS, sometimes TT, for saying that an event took place at some time. So, to express the proposition that `Harry hit Mary at time t,' one would write HOLDS(hit(harry,mary),t), where hit(harry,mary) is a functional expression denoting the event of `Harry hitting Mary.' The final and third approach, which has not been too popular in AI, is modal temporal logic. One introduces a number of modal operators (see Modal Logic), such as P (for past) and F (for future) and uses these to represent time-dependent information. For example, the information that `Harry hit Mary in the past' is represented as P(hit(harry,mary)). Reasoning about time is important in a number of areas in AI. In planning, for example, it is necessary to reason about the effects that an action will have on the world, and this involves reasoning about future states of affairs. Another example is natural language processing where one is concerned with extracting temporal information from the tenses of sentences.



Add Comment

No comments.