Program restructuring using agglomerative clustering technique based on binary features

Restructuring makes large programs highly cohesive and decomposes the statements into meaningful functions. These meaningful functions help the programmer to understand and maintain the code easily. In this paper, we apply agglomerative clustering technique to restructure the program using binary features. Our approach uses Jaccard similarity measure for binary features to assess the clustering process. We applied this approach to two non-cohesive structured programs available openly. We provide comparative evaluation in which clustering algorithms and similarity measures are used to validate our clustering results. Results indicate that this approach produces clustering that helps to translate a non-cohesive procedure into cohesive procedures.