Diachronic Description Logics

We provide motivation and a notation for representing in description logics certain \dynamic constraints" concerning changes of state. After providing a semantics for such descriptions using pairs of ordinary interpretations , we show that one can use existing DL inference engines to reason about such extended diachronic descriptions with no additional cost compared to the standard synchronic version. 1 Motivation Traditionally, concepts and individals in Description Logics describe single states of the domain of discourse. This is reeected even in the semantic interpetation function, which maps concepts to simple subsets of the domain of individuals. However, the real world and our knowledge of it is not static. Instead, actions (also called events or transactions) occur, changing one state to another. As part of describing an application domain, sometimes we would like to express constraints on the state changes. The following are some examples of such "dynamic integrity constraints", as these were called in the eld of databases a decade ago Lipeck&Saake 87]. 1. Unchanging properties/attributes. The property expressed in the real world is unchanging. Once you know the complete value(s) of the property for some individual in the knowledge base, this cannot change. For example, #birthMother for #PERSON /* knowledge of who gave birth to you doesn't change over time. */ 106