Reducing quantization error and contextual bias problems insoftware development processes by applying fuzzy logic

摘要

Object-oriented methods define a considerable number of rules,which are generally expressed using two-valued logic. For example, anentity in a requirement specification is either accepted or rejected asa class. There are two major problems of how rules are defined andapplied in current methods. Firstly, two-valued logic cannot effectivelyexpress the approximate and inexact nature of a typical softwaredevelopment process. Secondly, the influence of contextual factors onrules is generally not modeled explicitly. We term these problems asquantization error and contextual bias problems, respectively. To reducethese problems, we adopt fuzzy logic-based methodological rules. Thisapproach is method independent and is useful for evaluating andenhancing current methods. In addition, the use of fuzzy logic increasesthe adaptability and reusability of design models