Generalizing Updates: From Models to Programs

Recently the field of theory update has seen some improvement, in what concerns model updating, by allowing updates to be specified by so-called revision programs. The updating of theory models is governed by their update rules and also by inertia applied to those literals not directly affected by the update program. Though this is important, it remains necessary to tackle as well the updating of programs specifying theories. Some results have been obtained on the issue of updating a logic program which encodes a set of models, to obtain a new program whose models are the desired updates of the initial models. But here the program only plays the rôle of a means to encode the models. A logic program encodes much more than a set of models: it encodes knowledge in the form of the relationships between the elements of those models. In this paper we advocate that the principle of inertia is advantageously applied to the rules of the initial program rather than to the individual literals in a model. Indeed, we show how this concept of program update generalizes model or interpretation updates. Furthermore, it allows us to conceive what it is to update one program by another, a crucial notion for opening up a whole new range of applications concerning the evolution of knowledge bases. We will consider the updating of normal programs as well as these extended with explicit negation, under the stable semantics.