A New Approach for Requirements Management and Component Retrieval in A Cbd Based Software Development Using 2-Tuple Fuzzy Linguistic Approach

Abstract

Requirements management is a complex process in component based software development. A problem of requirements management is that requirements in general are incomplete, imprecise and contradictory. In an in-house development, the main objective is to implement a system, which will satisfy the requirements as far as possible within a specified framework of different constraints. In component based development, the fundamental approach is the reuse of existing components. The process of engineering requirements is much more complex as the possible candidate components usually lacking one or more features which meet the system requirements exactly. In addition, even if some components are individually well suited to the system, it is not necessary that they do not function optimally in combination with others in the system- or perhaps not at all. These constraints may require another approach in requirements engineering – an analysis of the feasibility of requirements in relation to the components available and the consequent modification of requirements. As there are many uncertainties in the process of component selection there is a need for a strategy for managing risks in the components selection and evolution process [31][32]. In this study, we are suggesting an efficient component retrieval system for a component search in a CBD based software development using 2-tuple fuzzy linguistic approach. Component retrieval is an activity that implies to achieve component that better fulfill the user requirements. For achieving this activity a component Retrieval System uses matching functions, which specify the degree of relevance of a component with respect to a user search for a component. Assuming linguistic weighted queries we present a new linguistic matching function, which is defined using a 2-tuple fuzzy linguistic approach [2] based on the functional description of the component. This new 2-tuple linguistic matching function can be interpreted as a tuning of that defined in [3] using an ordinal linguistic approach. We show that it simplifies the processes of computing in the retrieval activity, avoids requirements compromise in the final product, and consequently, can help to improve the users’ satisfaction and the quality of the product.