Extending the Cardinal Direction Calculus to a Temporal Dimension

Qualitative techniques for spatial reasoning are important in artificial intelligence. We present an extended cardinal direction calculus (XCDC) for spatio-temporal event representation and reasoning. The methods presented in this paper can be used in systems based on natural language processing which are also discussed in this paper. Introduction: Events as 3D entities A graph in a system of coordinates is a classical way of graphical representation of spatio-temporal relations. Similarly we consider an event as a combination of both a temporal span (time interval) and a spatial extension (space region). Although the time-space requires four dimensions to describe any possible object or event, we concentrate on events as spatio-temporal entities in a 3D space. This is because we analyse real-world situations as being located in flat open areas and we make abstraction from the actual altitude of these objects. The representation consists of two dimensions for space and one additional time dimension. The horizontal axes represent spatial aspects of an event. As it is necessary to choose directional orientation in CDC (see next section) these axes are denoted by N and W, meaning north and west respectively. The third axis denoted by T represents time.