An experience-based approach to software project management

Abstract

The paper describes an experience-based model for software project management. In this approach, the focus is on concrete cases or episodes, rather than on basic principles. Past problem solving experience is recorded as a collection of solution instances indexed using a set of labels. When a new problem is presented, the case-based reasoner searches its case repository and retrieves an old record that closely matches the current context. In the proposed scheme, we employ fuzzy sets to represent case indices and fuzzy aggregation functions to evaluate cases. This provides a formal scheme to quantify the partial matches of a given problem with multiple cases in the database and to utilize these partial matches to compute an aggregated result. Cost estimation and risk assessment functions of software project management are covered here. The paper discusses the basic principles of case-based reasoning and provides a functional description of the proposed system with details of case representation and case evaluation strategies.<>