BAYESIAN NETS: FROM LEGAL ARGUMENTS TO SOFTWARE RISK ASSESSMENT
Professor Norman Fenton, Centre for Software Reliability, City University

Abstract: Bayesian Belief Nets (BBNs) are an increasingly popular formalism for reasoning and decision-making under conditions of uncertainty. The underlying theory of BBNs is Bayesian probability theory. Although this has been around for a long time it is only in the last few years that efficient algorithms and tools have been developed to enable realistic sized BBN models to be executed. This talk will provide an introduction and overview of BBNs and will describe our own recent research in the subject. Specifically, I will describe our breakthrough methods for building large-scale BBNs and the novel applications where these have been used. I will describe applications in software risk assessment, where I believe we have addressed many of the weaknesses of classical software metrics approaches, but will also include a recent example of legal reasoning in criminal trials. The talk will include working demonstrations of BBNs in action. I will assume no previous knowledge of BBNs.

This seminar was held at the Department of Computer Science, Royal Holloway, University of London on 1 November, 1999.

back