An algebraic approach for granularity in qualitative time and space representation

Jérôme Euzenat (1)

1 SHERPA - Object Knowledge Bases

Any phenomenon can be seen under a more or less precise granularity, depending on the kind of details which are perceivable. This can be applied to time and space. A characteristic of abstract spaces such as the one used for representing time is their granularity independence, i.e. the fact that they have the same structure under different granularities. So, time "places" and their relationships can be seen under different granularities and they still behave like time places and relationships under each granularity. However, they do not remain exactly the same time places and relationships. Here is presented a pair of operators for converting (upward and downward) qualitative time relationships from one granularity to another. These operators are the only ones to satisfy a set of six constraints which characterize granularity changes. They are also shown to be useful for spatial relationships.

Domaines: Intelligence artificielle [cs.AI]

Jérôme Euzenat. An algebraic approach for granularity in qualitative time and space representation. 14th International Joint Conference on Artificial Intelligence (IJCAI), Aug 1995, Montréal, Canada. pp.894-900. ⟨hal-01401170⟩