Using Structured Knowledge Representation for Context-Sensitive Probabilistic Modeling

Abstract

We propose a context-sensitive probabilistic modeling system (COSMOS) that reasons about a complex, dynamic environment through a series of applications of smaller, knowledge-focused models representing contextually relevant information. COSMOS uses a failure-driven architecture to determine whether a context is supported, and consequently whether the current model remains applicable. The individual models are specified through sets of structured, hierarchically organized probabilistic logic statements using transfer functions that are then mapped into a representation supporting stochastic inferencing. We demonstrate COSMOS using data from a mechanical pump system.