Faikenhainer, B. and K.D. Forbus, Compositional modeling: finding the right model for the job, Artificial Intelligence 51 ( 1991 ) 95-143.

Abstract not found

Effective reasoning about complex engineered devices requires device models that are both adequate for the task and computationally efficient. This paper presents a method for constructing simple and adequate device models by selecting appropriate models for each of the device's components. Appropriate component models are determined by the context in which the device operates. We introduce context-dependent behaviors (CDBs), a component behavior model representation for encapsulating contextual modeling constraints. We show how CDBs are used in the model selection process by exploiting constraints from three sources: the structural and behavioral contexts of the components, and the expected behavior of the device. We describe an implemented program for selecting a simplest adequate model. The inputs are the structure of the device, the expected device behavior, and a library of CDBs. The output is a set of component CDBs forming a structurally and behaviorally cons...

Effective problem solving requires building adequate models that embody the simplifications, abstractions, and approximations that parsimoniously describe the relevant system phenomena for the task at hand. Compositional modeling is a framework for constructing adequate device models by composing model fragments selected from a model fragment library. While model selection using compositional modeling has been shown to be intractable, it is tractable when all model fragment approximations are causal approximations . This paper addresses the reasoning and knowledge representation issues that arise in building practical systems for constructing adequate device models that provide parsimonious causal explanations of how a device functions. We make four important contributions. First, we present a representation of class level descriptions of model fragments and their relationships. The representation yields a practical model fragment library organization that facilitates knowledge base co...

Many physical phenomena are sufficiently complex that the corresponding equations afford little insight, or no analytical method provides an exact solution. Decompositional modeling (DM) captures a modeler's tacit skill at solving nonlinear algebraic systems. DM divides statespace into a patchwork of simpler subregimes, called caricatures, each of which preserves only the dominant characteristics of that regime. It then solves the simpler nonlinear system and identifies its domain of validity. The varying patchwork reflects how variations in the parameters change the dominant characteristics. The patchwork is built by extracting equational features consisting of the relative strength of terms, and then exagerating and merging these features in different combinations, resulting in the different caricatural regimes. DM operates by providing strategic guidance to a pair of symbolic manipulation systems for qualitative sign and order of magnitude algebra. The approach is sufficient to re...

This report describes MM, a computer program that can model a variety of mechanical and fluid systems. It addresses several issues: What is the appropriate input to the modeling process? How should the search for models be organized? What evidence can be brought to bear to constrain the task? MM takes as input both a description of the structure of the system to be modeled and a trace of the system's behavior. The structure can be represented by 2D line drawings augmented with some additional information, or by component types and their interconnections, or both. The behavior trace provides the program with qualitative information about the behavior that the user wishes to capture with a model. MM organizes its search for models using a generalized, energy-based modeling framework and represents its models using bond graphs, a notation optimized for that framework. The program searches the system structure for plausible locations of primitive elements, connects the elements together ...