ISSN: 2320-5407 Int. J. Adv. Res. 5(9), 38-45 38 Journal Homepage: -www.journalijar.com Article DOI:10.21474/IJAR01/5298 DOI URL: http://dx.doi.org/10.21474/IJAR01/5298 RESEARCH ARTICLE EMPIRICAL APPROACH TO VALIDATE HEURISTIC ALGORITHMS FOR PROCEDURAL PROGRAMMING TO OBJECT ORIENTED DESIGN MIGRATION. Md. Samsuddoha 1 , * Md. Saeed Siddik 2 , Md. Selim 2 and Shah Mostafa Khaled 2 . 1. Department of Computer Science Engineering, University of Barisal. 2. Institute of Information Technology, University of Dhaka. …………………………………………………………………………………………………….... Manuscript Info Abstract ……………………. ……………………………………………………………… Manuscript History Received: 01 July 2017 Final Accepted: 03 August 2017 Published: September 2017 Key words:- Call Graph, Genetic Algorithm, Local Search, Variable Neighborhood Search, Build Cluster Hierarchy, Jaccard Similarity Coefficient. Although Object Oriented Programming is rapidly used in software industry, a wide variety of software is still used in the existing market developed in Procedural Programming languages. Sometimes, the maintenance of large software which developed in procedural languages become high costly and time consuming. To reduce this maintenance cost and make them more usable, some industries intend to transform that software from Procedural to Object Oriented Paradigm. On this purpose, researches have been focusing on automatic design migration approaches. Selecting a suitable and optimal method from the available is difficult because those approaches are not studied for any similar dataset. In this research, an empirical experiment has been conducted for identifying the optimal one among Genetic Algorithm, Local Search, Variable Neighborhood Search and Build Cluster Hierarchy algorithms. Six case studies from different real life software varying from 500 to 3700 LOCs are presented for experimental results verification. Final results are compared by a well known matrix named Jaccard Similarity Coefficient. Case studies show that Genetic Algorithm based approach outperforms other algorithms. Copy Right, IJAR, 2017,. All rights reserved. …………………………………………………………………………………………………….... Introduction:- Legacy software are still used in software industries and need regular up-gradation to keep pace with the market demands. Management of legacy software is not too easy to maintain and is very costly and time consuming. Most of this legacy software is developed in Procedural Programming (PP). On the other hand software developed in Object Oriented Programming (OOP) is very user friendly, usable and easy to maintenance. So, to reduce this maintenance cost and to make the legacy software more usable, several industries intend to transform this software from PP to OOP paradigm, which confirms some features like modularity, maintainability and manageability [1]. Sometimes these migrations are done manually, which is difficult and also time and resource consuming [2, 3]. Thus an automated migration from PP to OOP process is required. In the recent literature, several researches have done which addressed this design migration towards OOP paradigm. Most of those are based on graph clustering. Several heuristics algorithms are proposed for migration based on Monte Carlo and Greedy [4], Genetic Algorithm (GA) [5], Local Search (LS) [6], Variable Neighborhood Search (VNS) [7] and Build Cluster Hierarchy (BCH) [8]. An empirical analysis is required for measuring the performance and accuracy of these design migration algorithms. This research intends to make an empirical evaluation of design Corresponding Author:- Md. Saeed Siddik. Address:-Institute of Information Technology, University of Dhaka.