Contexts: Dynamic Identification of Common Parameters in Distributed Analysis of Complex Devices

Abstract

An important aspect of Distributed Problem Solving is that multiple intelligent agents understand each others communications. In our approach, domains (Thermodynamics, Mechanics, Fluids, etc.) are independendent problem solving modules that interact by resolving constraints on parameters common to each other. Since device behaviors change through time, these parameters appear and disappear through an analysis session. Further, each domain uses domain-specific languages for problem-solving efficiency and human usability. Hence, the communication paradigm must dynamically translate parameters from one domain to another with no a priori knowledge of what parameters may appear and what domains may be required during a session. Fortunately, parameters in applied physics carry a lot of spatial, temporal, and structural knowledge with them. In this paper we introduce the notion of contexts as an encapsulated description of this knowledge. We also show how contexts can be used efficiently to provide matching of parameters across domains.