Journal of Information Engineering and Applications                                            www.iiste.orgISSN 2224-5758...
Journal of Information Engineering and Applications                                               www.iiste.orgISSN 2224-5...
Journal of Information Engineering and Applications                                             www.iiste.orgISSN 2224-575...
Journal of Information Engineering and Applications                                            www.iiste.orgISSN 2224-5758...
Journal of Information Engineering and Applications                                                    www.iiste.orgISSN 2...
Journal of Information Engineering and Applications                                         www.iiste.orgISSN 2224-5758 (p...
Journal of Information Engineering and Applications                                                   www.iiste.orgISSN 22...
Journal of Information Engineering and Applications                            www.iiste.orgISSN 2224-5758 (print) ISSN 22...
Journal of Information Engineering and Applications                                         www.iiste.orgISSN 2224-5758 (p...
Journal of Information Engineering and Applications                                        www.iiste.orgISSN 2224-5758 (pr...
International Journals Call for PaperThe IISTE, a U.S. publisher, is currently hosting the academic journals listed below....
Upcoming SlideShare
Loading in …5
× modules clustering an effective approach for reusability


Published on

Published in: Technology, Education
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide modules clustering an effective approach for reusability

  1. 1. Journal of Information Engineering and Applications www.iiste.orgISSN 2224-5758 (print) ISSN 2224-896X (online)Vol 1, No.4, 2011 Software Modules Clustering: An Effective Approach for Reusability Jai Bhagwan1* Ashish Oberoi2 1. Department of Computer Science and Engineering, Maharishi Markandeshwar University, Mullana (Ambala) – 133 207, Haryana, India 2. Department of Computer Science and Engineering, Maharishi Markandeshwar University, Mullana (Ambala) – 133 207, Haryana, India * E-mail of the corresponding author: Phone: +91 98139 43210AbstractCluster analysis is a system used for cataloging of data in which data elements are screened into groupscalled clusters that represent collections of data elements that are based on a distance or dissimilarity. Thecluster analysis approach is an important tool in decision making and an effective creativity technique ingenerating ideas and obtaining solutions. This research covers different types of methods such asHierarchical Clustering (HC) and Non-Hierarchical Clustering (NHC) techniques for software modulesclassification.Keywords: Line of Code (LOC), Hierarchical Clustering (HC), Non-Hierarchical Clustering (NHC)1. IntroductionIn a knowledge engineering approach, the knowledge of human experts is described as a set of rules, whichare then used in the process of classification. The disadvantage of this approach is that it requires a lot ofefforts to make human knowledge explicit and for each new domain, again a separate formulation of therule need to be done manually. In a machine learning approach, the classifier is built without humanintrusion and classification for different domains can be learned using the same algorithm. The accuracy ofall automatic cataloging system is highly in need upon the effort and care taken during the rules definitionphase.A cluster analysis acts a big job in software alliance. Cluster analysis is the proposal for sorting out datainto clusters or groups in a situation where no prior information about a structure is vacant [17]. It dividesdata into groups (clusters) that are meaningful, useful or both [16]. The clustering approach is a key gadgetin decision making and an effective inspiration method in generating ideas and obtaining solutions. Thegoal of a cluster analysis is that units within a cluster should be as similar as possible, and clusters shouldbe as different as possible as shown in Figure 1. Clusters Objects Figure 1. Similar objects in similar clusters and others in different.18 | P a g
  2. 2. Journal of Information Engineering and Applications www.iiste.orgISSN 2224-5758 (print) ISSN 2224-896X (online)Vol 1, No.4, 2011Cluster analysis is a tentative data analysis tool for solving classified problems. Its objective is to sort cases(people, things, events, etc) into groups or clusters, so that the degree of friendship is strong betweenmembers of the same cluster and weak between members of different clusters [18]. Each cluster thusdescribes, in terms of the data composed, the class to which its members well in and this picture may beabstracted through the use of the particular to the general class or type. We have used HierarchicalClustering (HC) Algorithm and Non-Hierarchical Clustering (NHC) Algorithm in this research toimplement our proposed method described in coming topics.Cluster analysis is a course of identification and categorization of subsets of objects. Partitioning orclustering techniques are used in many areas for a wide spectrum of problems. Among the areas in whichcluster analysis is used are graph theory, business area analysis, information architecture, informationretrieval, resource allocation, image processing, software testing, chip design, pattern recognition,economics, statistics, biology and many more.2. Related WorkIn this section, the study of various papers and opinions of relative authors regarding the cluster analysis isprojected which aid in defining the objective. A brief picture of papers analysis is given under this topic.Manhas et al. in [1] discussed that the cost of software can be reduced by using software componentsalready existing, by their proper reuse. In this research various types of metrics are used to makeclassification of software modules. Back propagation based neural networks are experimented to obtain theresults in terms of Accuracy, MAE (Mean Absolute Error) and RMSE (Root Mean Squared Error values).Shri et al. in [2] propose that hybrid k-mean and decision tree approach which are used in order to find outthe reusability values of object oriented software components. The reusability values will be ‘Nil’ or‘Excellent’ in order to make distinguish between the components of software modules. The metrics forrecognizing the excellence of software components are predictable. The authors proposed an algorithm inwhich the data input is specified to k-mean clustering in the form of tuned values of the Object Orientedsoftware components. A decision tree system is developed in order to achieve similar type of softwareentities. The target of the metrics is to forecast the quality of software components. A range of attributeswhich are used in order to calculate the quality are maintainability, defect density, fault proneness,understandable, reusability etc. The purposed five metrics for OO Paradigms are as Weight methods perclass, Depth of inheritance tree, number of children, coupling between object classes and lack of cohesionin methods. The implementation is done in the open source tool named Rapidminer 4.6. On the basis of themetrics told above the result is calculated and shown by a decision tree using hybrid k-mean.Sembiring et al. in [3] analysed that there are various clustering methods in order to mine the data.Hierarchical clustering (agglomerative) is useful where clustering of variables is of agglomerative in nature.Hierarchical clustering is used in this research for mining the of record of an higher level learning institutein order to find out which course is suitable for students according to industries needs. MATLAB and HCE3.5 software is used to train data and cluster course applied during industrial training. This study showsdifferent methods form different number of clusters. Industrial training also used to acquire data to evaluatethe programme of the program. Continuous feedbacks from the industry and from the graduate studentsenable Higher Learning Institute to develop programme in order to suit industrial job needs. The methodslike hierarchical clustering method and divisive hierarchical method are used for data mining. TheAgglomerative method forms the cluster according to the similar and dissimilar entities. Dataset is takenfrom the parameter as questionnaires distributed during industrial training. In implementation 14universities courses, 17 faculty courses, 9 study program courses and 3 elective courses are taken. Theclustering is done on the basis of Euclidean distance and agglomerative methods.Author Sonnum et al. in [4] discussed that there are various kinds of methods for data finding and searchingwhich are successful. But they have one major lack that they don’t count the objects having not sufficientmatching properties. To overcome this problem we use Euclidean distance method. In this paper the19 | P a g
  3. 3. Journal of Information Engineering and Applications www.iiste.orgISSN 2224-5758 (print) ISSN 2224-896X (online)Vol 1, No.4, 2011Euclidean distance method with the Erlang language is used to implement a Web-based search on mobilephone application. Euclidean distance metric is the familiar distance between two objects and is given bythe Pythagorean formula. This research is done on the web database which is used to sell mobile phones.The sorting or clustering is done on the basis of their parameters like Mobile model, Resolution, width,height and weight. The data is mined by considering these parameters, so that a customer can easily pick upa mobile by his/her choice of quality. This research proves that objects having similar properties are easilyto find out with minimum time spending.Cancino et al. proposed that [5] clustering is done here to sort out the customer on the basis of theirelectricity consumed load, so Manila Electric Company will able to produce the electricity meter accordingto customers’ needs. The information can be obtained on the basis of the use of the daily load profile. Acase study is presented on the behalf of the actual load profile data sample residential customers ofdistribution utilities to determine the customer bracket that is affected by the existing Time of Use rates ofthe utility. The customer bracket was based on the energy consumption of all residential customers. Utilitieshave also been classifying its residential customers according to the KWh consumption use. This researchis done to analyze the appropriate clustering methods based on pattern recognition in load profiling forDistribution Utilities in a context of deregulated power industry. K-mean, Improved K-mean, agglomerativeand hybrid agglomerative methods are compared with.Czibula et al. [6] focused on the problem of decisive refactoring that can be used in order to improve thedesign of object oriented software systems. This paper aims at presenting a new hierarchical agglomerativeclustering algorithm, HARS (Hierarchical agglomerative clustering algorithm for restructuring softwaresystems). Clustering is used in order to overhaul the class structure of the system. HARS algorithm is usedin order to obtain an improved structure of a software system, by identifying the needed refactoring.Basit et al. in [7] focused that code clones are the software parts having similar or almost similar properties.Group of similar clones are known as structural clones or higher level clones. E.g. group of functions orclasses. The author found higher level clones using clone miner tool in order to manage design componentsof the software. We can easily understand the design of a software component. This approach followsformulation of structural clone and the application of data mining techniques to find out the higher levelsimilarity between clones. Clone miner finds simple clones first and then higher level or complex clones.The clone miner is written in C++ language. The author says this research is useful in softwaremaintenance.Sandhu et al. [8] discussed that software industries pay their attention to software reusability. Thesystematic use of the software reuse approach improves the productivity and quality of the software.However there are issues which have been restraining the wide broaden use of software reuse. This is allabout to the software components representation, its storage and rescue. Accepting and codifying theproperties of software components is vital to the useful management and development of component-basedsoftware systems. This paper gives a new classification idea for reusable software component based oninformation retrieval theory. Different organizations of the extracted keywords that represent the semanticfeature of the software component are evaluated. This approach allows using unwieldy terms and automaticcataloging of software components that are stored as a reusable component library. It could be beneficial forimproving the productivity of reuse repository manager by easy identification and retrieval of desiredsoftware components. The experiment is done in this research on 43 reusable software components whichare taken from ‘C’ based open access repositories. A program is developed in MATLAB 7.2 to test thevalidity of the proposed characterization scheme and model. The new model proposed model overcomessome of the limitations of the earlier models. The doublet and triplet word schemes make a better indexingunit as it also provides some contextual information as well. This is shown by the similarity function basedevaluation of the new model for grouping of the Software Components. Also it shows that the use of themodel will provide an effective characterization and indexing technique for storing out or grouping ofSoftware Components in Software Reuse Libraries. So, this approach can be used for automatic catalogingof software components and can be helpful in improving the productivity of the reuse repository managers.20 | P a g
  4. 4. Journal of Information Engineering and Applications www.iiste.orgISSN 2224-5758 (print) ISSN 2224-896X (online)Vol 1, No.4, 20113. Proposed ModelIn a large component repository, software component retrieval is accomplished through some classificationschemes [3]. The need of the software is increasing exponentially but the manpower is not increasingproportionately so we are in the situation of software failure or crisis. The reusability [4] is the excellenceof a portion of software that permits it to be used over again and again with or without a small change.Software professionals have indentified reuse as potent resources to potentially defeat the situation ofsoftware crisis [12] [13] [14]. The software module clustering problem consists of automatically finding agood quality clustering of software modules based on the relationships among the modules. Theserelationships typically take the form of dependencies between modules [15]. So we find relationshipbetween modules with LOC and Code Clones. As discussed earlier, we have used the existing HCalgorithm (Agglomerative method) [16] and NHC algorithm (K-Mean method) [17] in order to find out ourresearch objectives. For this we will follow our Algorithm as discussed below: 1. Search program modules from the system. 2. Find out LOCs and Clones from these modules. 3. Calculate Euclidian distance on the behalf of LOCs and Clones. 4. Find the no. of clusters to be made using Agglomerative method on the behalf of Euclidian distance. 5. Use K-Mean method to make clusters. 6. Calculate Efforts and Development time. 7. Compare Efforts and Development time after clustering process. 8. Graphical representation of results.The figure number 2 will make clear the picture for you, so you would be able to understand this approachmore easily.21 | P a g
  5. 5. Journal of Information Engineering and Applications www.iiste.orgISSN 2224-5758 (print) ISSN 2224-896X (online)Vol 1, No.4, 2011 Find LOC and Clones from these modules. Calculate Euclidian distance on Find modules from the the behalf of LOCs and Clones. system (computer storage). Find the no. of cluster to be made using Agglomerative method. Graphical representation of Use K-mean method to make results clusters. Compare efforts and development Calculate efforts and time after clustering process. development time. Figure 2. The proposed model given by our self.4. Results DiscussionWe have experimented on 15 C++ modules. After attaching the codes to JB Clone Scanner (developed inC#), we find the results in term of LOC of each module and clones between each module as shown in table1. Using Agglomerative [3] method we find the schedule on the behalf of which we decide how manyclusters we can make. This is shown in the table 2. You see there is a big jump in coefficient values afterstage 6. So the number of cluster will be 9 because we deduct 6 from 15 i.e. 9, after applying the equation 1given below: Number of Clusters = N - Si (1)  Where N is the total number of modules.  Si is the case stage up to where the coefficient values difference is smaller to the next stages. Table 1. LOC and Clones of each module Module Line of Code (LOC) Clones 1 43 20 2 29 21 3 29 1622 | P a g
  6. 6. Journal of Information Engineering and Applications www.iiste.orgISSN 2224-5758 (print) ISSN 2224-896X (online)Vol 1, No.4, 2011 4 47 23 5 30 14 6 21 11 7 21 12 8 18 9 9 62 18 10 23 12 11 17 11 12 20 19 13 22 9 14 15 6 15 15 7After attaching the Lines of Code of each module to COCOMO Basic tool (developed in C#) we find thedifferent results before and after clustering approach. The comparisons of these results are shown with thehelp of figure 3, 4 and 5.Table 3 shows how the LOC and Clones are adjusted in total 9 clusters. Table 2. Agglomerative Schedule using average linkage method Stage Cluster Combined Coefficients Cluster 1 Cluster 2 1 14 15 1.000 2 6 7 1.000 3 6 10 4.500 4 8 11 5.000 5 3 5 5.000 6 6 13 8.333 7 8 14 20.000 8 1 4 25.00023 | P a g
  7. 7. Journal of Information Engineering and Applications www.iiste.orgISSN 2224-5758 (print) ISSN 2224-896X (online)Vol 1, No.4, 2011 9 2 3 37.500 10 6 8 45.375 11 2 12 100.000 12 2 6 163.000 13 1 9 307.500 14 1 2 1022.194 Table 3: Final Cluster Centre Cluster 1 2 3 4 5 6 7 8 9 LOC 43 29 30 47 15 20 22 18 62 Clones 20 21 15 23 7 19 11 10 18 Lines of Code before and After Clustering 500 400 300 200 100 0 Before After Clustering Clustering Figure 3. Comparison of Lines of Code24 | P a g
  8. 8. Journal of Information Engineering and Applications www.iiste.orgISSN 2224-5758 (print) ISSN 2224-896X (online)Vol 1, No.4, 2011 Efforts before and After Clustering 1200 1000 800 600 400 200 0 Figure 4. Comparison of Development Efforts. Development Time before and After Clustering 200 150 100 50 0 Figure 5. Comparison of Development Time.25 | P a g
  9. 9. Journal of Information Engineering and Applications www.iiste.orgISSN 2224-5758 (print) ISSN 2224-896X (online)Vol 1, No.4, 20115. ConclusionA large numbers of software are available in all over the world and the need of software increasing day byday but the manpower is not increasing proportionately. So there is a big need to reuse those softwaremodules in new projects. For this intention it is very crucial to organize those software modules in ameaningful structure, which can be done by cluster analysis. After applying clustering methods, we find thenatural groups of software module where the degree of similarity between two software modules ismaximal if they belong to the same group and minimal otherwise and we also find the maximum similarclones between software modules within similar cluster. We saw that after implementation of clusteranalysis number of LOCs, efforts of development and development time are reduced. So this research iseffective to overcome the problem of software crisis or failure. This experiment has done with 15 C++modules only. This approach can be effective for a large dataset and automation of this approach will be amilestone in software engineering.References[1] Manhas S., Sandhu P. S., Chopra V., Neeru N. (2010), “Identification of Reusable Software Modules in Funcion Oriented Software System using Neural Network Based Technique”, World Academy of Science, Engineering and Technology, Vol. 67.[2] Shri A., Sandhu P. S., Gupta V., Anand S. (2010), “Prediction of Reusability of Objected Oriented Software System using Clustering Approach”, World Academy of Science, Engineering and Technology, Vol. 67, PP. 853-856.[3] Sembiring R. W., Zain J. M., Embong A. (2010), “A Comparative Agglomerative Hierarchical Clustering Method to Cluster Implemented Course”, Journal of Computing, Vol. 2, Issue 12, PP. 1-4.[4] Sonnum S., Thaithieng S., Ano S., Kusolchu K., Kerdprasop N. (2011), “Approximate Web Database Search Based on Euclidean Distance Measurement”, Proceeding of the International MultiConference of Engineers and Computer Scientists, Vol. 1, PP. 16-18.[5] Cancino A. E. “Load Profiling of MERALCO Residential Electricity Consumers using Clustering”, Manila Electric Company (MERALCO), Pasig City, Philippines.[6] Czibula I. G., Serban G. (2007), “Hierarchical Clustering for Software System Restructuring”, Babes Bolyai University, Romania.[7] Basit H. A., Jarzabek S. (2007), “A Data Mining Approach for Detecting Higher-level Clones in Software”, IEEE Transactions on Software Engineering, PP. 1-18.[8] Sandhu P. S., Singh H., Saini B. (2007), “A New Categorization Scheme of Reusable Software Components”, International Journal of Computer Science and Network Security, Vol. 7, Issue 8, PP. 220-225.[9] Maqbool O., Babri H. A., “The Weighted Combined Algorith: A Linkage Algorithm for Software Clustering”, Lahore University of Management Sciences DHA Lahore, Pakistan.[10] Nakkrasae S., Sophatsathit P., “A Formal Appraoch for Specification and Classification of Software Components”, University of Louisiana at Lafayette, U.S.A., Chulalongkorn University, Bangkok, Thailand.[11] Jalender B., Govardhan A., Premchand P. (2011), “Breaking the Boundaries for Software Components Reuse Technology”, International Journal of Computer Applications, Vol. 13, Issue 6.[12] Goel H., Singh G. (2010), “Evaluation of Expectation Maximization based Clustering Approach for Reusability Prediction of Function based Software System”, International Journal of Computer Applications, Vol. 8, Issue 13.26 | P a g
  10. 10. Journal of Information Engineering and Applications www.iiste.orgISSN 2224-5758 (print) ISSN 2224-896X (online)Vol 1, No.4, 2011[13] Jeng J. J., Cheng B. H. C. (1993), “Using Formal Methods to Construct a Software Component Library”, Proc. of 4th Eur. Soft. Eng. Conf., Lect Notes in Comp. Science, PP. 397-417.[14] Jeng J. J., Cheng B. H. C. (1995), “Specification Matching for Software Reuse: A Foundation”, ACM, PP. 97-105.[15] Mahdavi K., Harman M., Hierons R. M. (2003), “A Multiple Hill Climbing Approach to Software Module Clustering”, Proceedings of the International Conference on Software Maintenance, DISC Brunel University.[16] Fokaefs M., Tsantalis N., Chatzigeorgiou A., Sander J. (2009), “Decomposing Object-Oriented Class Modules Using an Agglomerative Clustering Technique”, IEEE, Proc. ICSM, Canada.[17] Kanungo T., Mount D. M., Netanyahu N. S., Piatko C. D., Silverman Wu A. Y. (2002), “An Efficient k-Means Clustering Algorithm: Analysis and Implementation”, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 24, Issue 7.First Author, Jai Bhagwan has been a bonafide student of Master of Technology (Computer Science andEngineering) degree in Maharishi Markandeshwar University, Mullana, India during the period 2009-2011.He also worked as a software developer in a government - private initiative for 7 months before his M.TECHdegree.He had been awarded for Best Explained Model of PSLV satellite in Kurukshetra University, Kurukshetraduring his schooling. His current research interests are Software Engineering and Data Mining. Currentlyhe is working as Lecturer/Assistant Professor in IT department of Maharishi Markandeshwar University,Mullana, India.Second Author, Ashish Oberoi is working as Lecturer/Assistant Professor in Computer Science andEngineering department of Maharishi Markandeshwar University, Mullana, India. He is Master ofTechnology in CSE discipline. He has 6 years of teaching experience and he is pursuing his PhD in ImageProcessing field.His research area is Image Processing and Software Engineering.27 | P a g
  11. 11. International Journals Call for PaperThe IISTE, a U.S. publisher, is currently hosting the academic journals listed below. The peer review process of the following journalsusually takes LESS THAN 14 business days and IISTE usually publishes a qualified article within 30 days. Authors shouldsend their full paper to the following email address. More information can be found in the IISTE website : www.iiste.orgBusiness, Economics, Finance and Management PAPER SUBMISSION EMAILEuropean Journal of Business and Management EJBM@iiste.orgResearch Journal of Finance and Accounting RJFA@iiste.orgJournal of Economics and Sustainable Development JESD@iiste.orgInformation and Knowledge Management IKM@iiste.orgDeveloping Country Studies DCS@iiste.orgIndustrial Engineering Letters IEL@iiste.orgPhysical Sciences, Mathematics and Chemistry PAPER SUBMISSION EMAILJournal of Natural Sciences Research JNSR@iiste.orgChemistry and Materials Research CMR@iiste.orgMathematical Theory and Modeling MTM@iiste.orgAdvances in Physics Theories and Applications APTA@iiste.orgChemical and Process Engineering Research CPER@iiste.orgEngineering, Technology and Systems PAPER SUBMISSION EMAILComputer Engineering and Intelligent Systems CEIS@iiste.orgInnovative Systems Design and Engineering ISDE@iiste.orgJournal of Energy Technologies and Policy JETP@iiste.orgInformation and Knowledge Management IKM@iiste.orgControl Theory and Informatics CTI@iiste.orgJournal of Information Engineering and Applications JIEA@iiste.orgIndustrial Engineering Letters IEL@iiste.orgNetwork and Complex Systems NCS@iiste.orgEnvironment, Civil, Materials Sciences PAPER SUBMISSION EMAILJournal of Environment and Earth Science JEES@iiste.orgCivil and Environmental Research CER@iiste.orgJournal of Natural Sciences Research JNSR@iiste.orgCivil and Environmental Research CER@iiste.orgLife Science, Food and Medical Sciences PAPER SUBMISSION EMAILJournal of Natural Sciences Research JNSR@iiste.orgJournal of Biology, Agriculture and Healthcare JBAH@iiste.orgFood Science and Quality Management FSQM@iiste.orgChemistry and Materials Research CMR@iiste.orgEducation, and other Social Sciences PAPER SUBMISSION EMAILJournal of Education and Practice JEP@iiste.orgJournal of Law, Policy and Globalization Global knowledge sharing:New Media and Mass Communication EBSCO, Index Copernicus, UlrichsJournal of Energy Technologies and Policy Periodicals Directory, JournalTOCS, PKPHistorical Research Letter Open Archives Harvester, Bielefeld Academic Search Engine, ElektronischePublic Policy and Administration Research Zeitschriftenbibliothek EZB, Open J-Gate,International Affairs and Global Strategy OCLC WorldCat, Universe Digtial Library ,Research on Humanities and Social Sciences NewJour, Google Scholar.Developing Country Studies IISTE is member of CrossRef. All journalsArts and Design Studies have high IC Impact Factor Values (ICV).