Merging model driven and ontology driven system development approaches pervasive computing perspective

Abstract

In this paper we present a view point on ldquointelligentrdquo application development for pervasive computing environments. We first point out that today's traditional ldquointelligentrdquo computing is built on strong and hard-coded logical assumptions and computational procedures which are pre-defined by developers, that is, what we call as simulated intelligence within the course of this paper. Such assumptions and procedures are based on enumerations of possible contexts of use which is predefined mappings between situations in contextual space to rational behaviors in behavior space. However pervasive computing applications extend the scope of application's context space and behavior space towards infinity which hardens development of ldquointelligentrdquo systems having a certain degree of rationality. Therefore, approaches merging human intelligence and computing ldquointelligencerdquo are required to be employed. We further advocate that pervasive computing era increases the complexity of application development because of the extended context space, hence software development approaches based on higher abstractions need to be employed where model driven approaches and ontology driven approaches are promising. We propose a basic methodology which merges model driven and ontology driven development approaches. Resulting methodology employs use of formalized conceptual models to be employed both at run-time and development time in terms of reasoning and automatic code generation respectively. We finally point out that such application development paradigm based on pervasive computing perspective will enable users to program their own environment (i.e. smart spaces) in the future.