Reusing Components using Genetic Algorithms

Abstract

The need to improve software productivity and software quality has put forward the research on software reuse technology. Today’s, complex, high quality computer based systems must be built within budget and time schedule. This demand leads to reuse of existing components. Software Reuse involves four steps definition, retrieval, adaptation and incorporation. A structured software library, where software components are properly classified is required so that the user can easily retrieve the software components. The main technical problems, which limit the practice of reuse, are effective structuring of the software repository and a mechanism to effectively retrieve from the repository. Structuring library of software components represented by natural languages may hinder the retrieval process due to the problem of ambiguity, incompleteness and inconsistency inherent to natural language. Formal Specifications, representing the software components are unambiguous, complete, and consistent and can be verified. The approach presented in this thesis uses Genetic Algorithms with Formal Specifications for clustering the software components, in which the number of clusters formed depends upon the similarity in cluster and similarity between clusters. The aim is to find optimized clusters in which components are classified. For cluster based retrieval the main aim is to find the cluster containing components similar to the given query.