HP Labs Technical Reports
Click here for full text:
An Adaptive Model for Explicit Uncertainty Management
Piccinelli Giacomo; Casassa Mont, Marco
Keyword(s): CBR; case-based reasoning; adaptivity; information retrieval; uncertainty
Abstract: Uncertainty is a peculiar aspect of the knowledge gathered by a diagnostic system. Looking at content, relevance and confidence as the basic elements of system knowledge, we notice that content (facts and rules) and relevance are widely exploited into diagnostic models and tools but confidence doesn't usually receive specific attention. We present a model in which uncertainty becomes a fundamental and dynamic component of both diagnostic knowledge and processes: fuzzy sets are the theoretic base of the model. A conversational shell has been developed in order to test the impact of our proposal on system performance and we discuss both short-term and long-term benefits of explicit uncertainty management.
Back to Index