1 RULE-BASED BEHAVIOURAL MODELLING OF INFORMATION SYSTEMS A. Tsalgatidou, D. Gouscos, C. Halatsis Dept. of Informatics, University of Athens, TYPA Buildings, Panepistimiopolis, Ilisia, GR-157 71 Athens, Greece email: afrodi1@grathun1.bitnet 1. Introduction The development of information systems is a stepwise process which, at the very early stages, requires a comprehensive specification of the features to be developed. These features are of two kinds: features representing the static system aspects, i.e. structural relationships, and features representing the dynamic system aspects, i.e. aspects that define the system behaviour. The modelling of dynamic system aspects is of special importance as it expresses the application domain policy defining in this way the behaviour of the information system to be developed. Changes to application domain policy must be directly reflected to information system aspects, as these require respective changes to be performed in the information system model, so that it can reflect the current state of the application being modelled. Thus, the explicit modelling of dynamic system aspects is one of paramount importance for the development of flexible information systems, i.e. systems that can react to changes of the external environment. This paper focusses on the modelling of the dynamic system aspects, which define the behaviour of information systems, though it covers the static aspects too. To this end, it introduces a rule-based model capable of modelling dynamic system aspects within an object-oriented framework. More specifically, in the approach proposed in this paper, the capturing and definition of the static properties and the relationships between the entities identified in an information system, is achieved by employing an entity-relationship modelling technique enriched with inheritance, on the basis of a linguistic approach [12], [17]. In the sequel, the derived binary entity- relationship model is mapped to an object-oriented model. This object-oriented model represents the specification of the static aspects of an information system. The dynamic aspects of a system are specified by the dynamic aspects of each identified object class. In the proposed approach the dynamic behaviour of each object class is described by a set of dynamic rules triggered by specific signals. Dynamic rules triggered by the same signal are grouped in one Behaviour Unit (BU). Thus, the behaviour of each object class is described by a set of Behaviour Units triggered by a specific signal, while the behaviour of each BU is described by a set of dynamic rules. Signals are generated either by the external environment or by the dynamic rules and are directed either to the external environment or to specific parts of the modelled system. In this way, signals constitute the means of communication between the various parts of an information system and between the system and its external environment. At this stage, the constructed object-oriented model covers both static and dynamic aspects of the system under study. Furthermore, a Petri net [14] based model, called Rule-Based Net (RBN), is automatically derived from the constructed object-oriented model and is used for the graphical representation of the dynamic aspects. An RBN is a Petri net like model, augmented with signals and rules. In the context of RBNs, Petri net places represent