Extracting and Organizing Software Elements via Intelligent Classification

Abstract

An innovative approach to intelligent software component categorization and retrieval according to user-defined needs is presented
in this study. By partitioning the pool of accessible components kept in a database into specific subsets (clusters), the
classification method makes use of a specialized genetic algorithm to develop a limited set of classifiers. As a result, each classifier
takes on the role of cluster leader. The user specifies the required qualities (component profile) while tracing a component,
and these are compared to the available classifiers’ features. The user is shown the “winning” set of components from each
cluster in decreasing matching fitness, where the closest classifier matching the necessary criteria above a user-defined threshold
is the outcome. We have successfully tested our methods on a simulated dataset of components, and the results are promising.
Finally, we showcase the web app that was made to back up the intelligent categorization approach that was suggested.