This paper presents a method for modeling temporal scenarios, and for recognizing them from observations. Three formal models are developed: the logical dependencies, defined in possibilistic logic; the temporal constraints between events, managed as communicating sequential processes (CSP); and a third tree structure that matches observations and scenarios.

The situations modeled are typical of dynamic scenarios where a strategy or tactic exists, namely defense and surveillance systems. The formal model applies some simplifications, for example assuming the existence of only one scenario at a time.

While the idea of separating logical dependencies and temporal reasoning is not new, the most interesting aspect of this method is its matching between observations and scenarios, obtained through an aggregation tree, that introduces a weight, inspired by fuzzy logic, to express the degree of matching.

The developed model will be of interest in the broad area of pattern recognition for structural information, an area of intersection between pattern recognition and artificial intelligence. The model is still too simple to accommodate real-life situations; however, it proposes a way to integrate structural information and reasoning.