Comparative Study on DFD to UML Diagrams Transformations

World of Computer Science and Information Technology Journal(WCSIT) ISSN: 2221-0741 Vol. 1, N o. 1, 10 -16, 20 11. 10 Comparative Study on DFD to UML Diagrams Transformations Atif A. A. Jilani Muhamm ad Usman Aamer Nade em National University of Co mputer Mohammad Ali J innah University Islamabad Mohammad Ali Jinnah University Islamabad and Emerging Sciences , Islamabad atif.jilani@nu.edu.p k m_usman99@ hotmail.com anad eem@jinnah.edu.pk Zafar I . Malik Zahid H alim Academy of Education & Plannin g, Ministry of National Universit y of C omputer and Emerging Education, Pakistan Sciences, Islamabad za farimalik@hotmail.com zahid.halim@nu.edu.pk Abstract — Most of lega cy system s use n owadays w ere modeled an d docum ented usin g structu red appr oach. Expan sion of th ese system s in terms o f functionality and maintainability requires shift towards o bject-oriente d d ocum entation and design , which has b een widely accepted by the industry. In this paper, we present a survey o f the existing Data Flow Diagram (DFD) to Unified Modeli ng language (UML) transform ation techniques. We analyze transf ormation techn iques using a set of paramete rs, identifie d in the s urvey. Based on i dentifie d paramete rs, we pr esent an an alysis matr ix, w hich describes th e stren gths and w eaknesses of trans forma tion techniques. It is observ ed that most of the transform ation approaches are rule base d , which are incomplete and define d at abstract leve l that does not co ver in depth transf ormation and automation issu es. Transformation appr oaches are data centric, whi ch focuses on data- store for class diagram generation. Very fe w of the transformation techniqu es have been applied on case study as a proof of c oncept , which ar e not com prehensive and majority of them are parti ally autom ated. Keyw ords -Unif ied Modeling Langu age (UML); Data Flow Diagram (DFD); Class Diagram ; Model Transform ation. I. I NTRODUCTI ON Most of the legacy systems in use now adays were modele d and documented using structu red approach [1]. T hese system s were developed in languages that have become o ut-dat ed now . With the pass age of time, sy stems demand num erous modificat ions, expansion in terms o f functionality and incorpora tion of latest hig h-speed hardw are. Still, leg acy system s are reli able enoug h and consid ered ir replaceable by the user. However, it is possible to modify s ys tem code but modificat ions in the code add inconsisten cies between code and design an d system design becomes no long er usable f or future maintenance . Besides , modifying such a system is also very costly, the only viable solution for up-gradati on and mainten ance is to preserve system design and in corporate it with latest softw are developm ent strategies as described by Newcom be and Doblar [2 ]. If running system code is available , it is possi ble to generate d esig n from code. H owever, if the code is modified numerous times, the generated design and origina l design may become inconsistent. Desig n r ec overy from such code is ambiguou s an d no more u seful for future up-gradation and mainten ance. Like Dietrich et al. [3], we also consider legacy system s irreplaceabl e and tr usted by the users. We, too, emphasize on saving legacy system by providing and using an object-ori ented inte rface. A major design artifa ct in stru ctured approach is the Data Flow Diagram (DFD). Other artifacts like structure chart, state machin e, and ER diagram are also there, but DFD has certain advantages over them . DFD is the prim ary artifact and is required be created for every systems in structure d approach . DFD has hierarc hal structure, which provides different abstracti on level, us eful in sy stem desig ning. Besides, DFD is such a fundam ental artifact that clearly depicts the stru cture o f a sy stem. Other artifacts use the inform ation provided b y the DFD to represent dynamic aspect of the sy stem [3]. Structure d design techniques h ave been replaced by object - oriented analy sis and d esign approach, which has gained popularity n ow and majority of the s oftware modeling and developm ent techniques are adopting this p arad igm [4]. W ith the passage o f tim e, the level of abstraction in sy stem developm ent has raised. Obje ct Management Group (OMG) has been recently prom oting a new vision for softw are developm ent, i.e., Model Driven Architecture (MDA) [5]. In MDA, main emphasis is on modeling design separately from the implementati on (platform). MDA encourages the use of Platform Independ ent Model (PIM) and Platform Specif ic Model (PSM) to represent p latf orm independent and platfor m specific details. Soul of MDA is tr ansfo rmation between models. In MDA, code may be generated either from PSM or from the PIM . Model transf orm ations can be between PIM and PSM, PIM and code or betw een PSM and code. For modeli ng object-ori ented systems and model creation, Unified Modeling Languag e ( UML) [6] has no w beco m e the de -fact o indus try standard [7 -8]. UM L is a colle ction of diagram s used to model WCSIT 1 (1), 10 -16, 2011 11 different aspects of object orient ed software. UML Class diagram is one of a major artifact in object - oriented desi gn used to represent the system’s static stru cture. Other U ML diagram s, like sequence diagram, state machine, and activity diagram etc, are us ed to m odel the sy stem’s dy namic behavio r. In this paper, w e present a survey of transformation techniques that are used to generate legacy s ystem d esign in UML. We include DFD to UML diagrams transform ati on in our survey. We analyze different existing transformation techniques using set of Analy sis param eters identifie d in the survey . Based o n the parameters analysis matrix is created, which highlig hts the weaknesses and strengths of different techniques. Motivation behind DFD-UML models transform ations is that design ers/analys ts can use surveyed transform ations from DFD to class diagram, with the existin g MDA transformation [9] either as PIM to PSM or as PIM to code. II . S COPE OF THE S URVEY Modernizati on of legacy system s cost effectively has become the primary focus of softw are designers and researche rs. In literatu re, both structur ed code to object- oriented design and structu red design to object-oriente d design transform ations exists . Code to design techniques cause mainten ance issue b ecause code is written in programm ing languages, wh ich have become o ut-date d now. Besides, code might have undergone numerous modifi cations, which make code and design inconsistent . Design recovery from that code is useless an d no more usable for futu re maintenanc e. Contrary to this, in structu red desig n to object-orient ed desig n transform ations, new d esign provid es basis for development of new s ys tem and b ecomes future reference for maintenanc e. We limit our scope to legacy system design to UML design. In particula r, we focus o nly on data flow diagram to UML design transform ation. In literatur e, both structu red code to object-oriented design and structure d design to non-UML o bject-or iented design techniques exists. In this section, w e discuss both the view s. A. Structured Code to Object-o riented Design Techniques In structu red code to object-oriente d design techniques Liu and Wilde [9 ], propose methodologies for identif ying object from non-object-oriented languages. T hey propose type ba se and glo bal base object fin der m ethodologies. Jacobson and Lindstrom [10] describe reverse engineering strategi es and discuss object-orien ted model to incorporate changes. Livadas and Jo hns on [11] propose an approach that maintains existing relationsh ip in the m aintained code. Simila rly, another approach by Gall and Klösch [12] defines two types of data entities: data store entiti es and non-data store entities . They describe relati onship betw een the two types for expr essing entities as objects . Newcom be and Kotik [13] present a tool for abstract object-ori ented model genera tion. Subram anian and Bwirne [14] generate objects from FORTRAN code. They discuss constrain ts like private, virtual, and pure v irtual . Cimitile et al [15] and De Lucia et al [16] present approaches that revolve around data stores. Authors propose approaches that consid er functions an d subroutines, int eracting with tables, data -st ore and use them as objects methods . Similarly , De Luci a et al [17] propose an approach that recover cla ss diagram from data intensive legacy sy stem code. It is apparent that many of the discusse d techniques are effective only for data centric systems. For our approach, we are firm ly interested in stru ctured d esign instead of co de. In design, we have observed that in lit erature, both stru ctured design to non-UML design and structu red design to UML design trans formations exist. We briefly explain both the view s. B. Structured Design (DFD) to a Non UML Object- oriented Design In this section, we will discuss those techniques that transform structured design to a non -UML object-orient ed design. Alabis o [18] use FDC ( Functi onal Design Chart ) to express fun ctional behavi ors and OSC (Object Structure Chart) to express breakdown of data structures of object. His transform ation approach is not automatable and does not provide detail transformati on rules . Ge orge an d Ca rter [19 ] propose mapping strategy that uses Entity Relationship Diagram (ERD), Functional Data Flow Diagram (FDFD) and Data Dicti onary as a source model and generates Objec t Structure and Mapping Diagra m (OSMD). Their approach too is not au tomatable. III . C RIT ERIA FOR E VALUATI ON In this section, w e descri be and discuss s et of analy sis paramete rs. Analy sis par amet ers provide criterion to evaluate different techniques . For parameters sele ction, we comprehensively consider transform ation techniqu es description , their limitations and comparis on discu ssed by the auth ors w hile describing their respective technique. On review ing d ifferent techniques, certain parameters are iden tified. Detailed description of paramete rs and thei r values is given b elow. A. Automatable This parameter describes whether a transfo rmation technique is automated or can only be applied manu ally. The paramete r value depicts practical importance of t he techniq ue and used in determ ining the efficiency and applica bility of the approach . The values assigned to the param eter are ‘Yes’, ‘No’ and ‘ Partial’. Table 1 shows value selecti on criter ia for automata ble. T ABLE 1 E VALUATION CRITERIA F OR A UTOMATION . Value Criteria Yes Technique automatical ly transform source model into target model. No Technique does not automatically transform source model into target model. Partial Author expl icit ly mentioned or after analysis of case study and transformation rules, we have found that some manual su pport is needed for automation. B. Tool Support Tool support parameter describes w hether a tool or an automata ble environment is available for the technique or not. Table 2 sh ows v alue selecti on criteria for tool su pport. T ABLE 2 E VALUATION CRITERIA F OR T OOL S UPPORT . Value Criteria Yes Author expl icitly mentioned tool support. No No information re lated to tool is prov ided. WCSIT 1 (1), 10 -16, 2011 12 C. Add itional Artifact Used This param eter describ es the addit ional artif acts used by the transform ation technique. Values for this parameter include Data Dictiona ry (DD), E nti ty Relationship Diagra m (ER) etc . D. Outp ut Artifact This parameter describes UML artifacts that are genera ted because of transfo rmation technique . Values fo r this param eter could be Cl ass, Use- Case, Sequen ce, state cha rt diag ram etc. E. Case Study This parameter defines wh ether a transform ation techniqu e applied on a case study or n ot. This parameter is import ant because a techni que needs to be appl ied on case stu dy to check it applicability . T able 3 shows value selection criteria for case study. T ABLE 3 E VALUATION CRITERIA F OR C ASE S TUDY . Value Criteria Yes Case Study prov ided or discussed. No No Case study discussed. F. Transformation level This parameter defines type of transform ation given technique follow s. Value for this parameter includes rule- based or metamod el-based transform ation. Table 4 show s value sele ction crite ria f or transf ormation lev el. T ABLE 4 E VALUATION CRITERIA F OR T RANSFORMATION L EVEL . Value Criteria Rule-based Transformation rul es are provided for models instances. Metamodel-based Transformation mappings and rules are provided at metamodel l evel. G. UML Confo rmance This Parameter describ es the conf ormance of the generated model. I t shows model produce by transformation technique follow s UML syntax and sem antics or not. Table 5 show s value sele ction crite ria f or UML confo rmance. T ABLE 5 E VALUATION CRITERIA F OR UML C ONFORM ANCE . Value Criteria Yes Generated model follows UML syntax and semantics, describe in UML Superstru cture. No Generated mode l do not foll ow UML syntax and semant ics. H. S calability This parameter defines wh ether a transform ation techniqu e has potential, to be applied on larger size case study or not. Case study descripti on provides the basis for this parameter. Table 6 sh ows v alue selecti on criteria for scala bility. T ABLE 6 E VALUATION CRITERIA F OR S CALABILITY . Value Criteria Yes Size of case st udy is appropriate and technique fully transformed source model to t arget model in feasible time. Technique de scription helps in de termine in efficienc ies. No Size of case stud y is small and approach does not fully transformed mo del in feasible time. I. Direction This p aram eter defin es transform ation direct ion. Values include unidirecti onal and bidirectional . T abl e 7 shows value selection criteri a for direc tion. T ABLE 7 E VALUATION CRITERIA F OR D IRECTION . Value Criteria Bidirectional Technique transforms structured model to object -oriented model and can retransform object-oriented model to structured model . Unidirectional Technique transforms structured model to object -oriented model only . J. Input Model Covera ge Input model coverage parameter describes the coverag e of different constru cts of source m odel. Since, our primary focus is on DFD, we check that different constru cts of DFD like external ent ities, data- stores, pro cesses and dat a-flows betw een them are properly transform ed into target model or not. Table 8 show s value selec tion crite ria for in put mod el coverage . T ABLE 8 E VALUATION CRITERIA F OR I N PUT M ODEL C OVERAGE . Value Criteria Low Basic constructs like external entities, d ata-stores, p rocesse s are only transfor med. Medium Few data-flows between components along with basic constructs are tr ansformed High All data -flo ws between components along with all th e constructs are tr ansformed K. Target Mo del Coverage: Target model cov erage parameter describes the coverage of different constru cts of the target model. Since, we focus o n UML m odels this param eter descri be constru ct of d iffer ent UML models. For class diagram , co nstruc t o f class diagram like owned attribute, owned operation, associa tion, dependency , inheri tance relationsh ip are co nsi dered. Class diagram metamodel describe in UML superstructure, helps in identify ing different construc ts. Table 9 shows value selection criteria for output model coverage. T ABLE 9 E VALUATION CRITERIA F OR T ARGET M ODEL C OVERAGE . Value Criteria Low Basic constructs, like owned attributes and owned operation are catered. Medium Association between components along with basic constructs is catered. High All the constr ucts including inherit ance are catere d. L. DFD Level Used: This parameter defines the level of DFD used in transform ation t echnique. Values for this paramete r inclu de context level DFD and Extended level DFD. Table 10 shows value sele ction crite ria f or DFD level used. T ABLE 10 E VALUATION CRITERIA FOR DFD L EVEL U SED . Value Criteria Context Le vel DFD in transformation has hierarchy which is n ot refined Extended Le vel DFD in transformation is refined and no hierarchy exists. Based on the above-m entioned c riteria, we evaluate th e techniques explained below . IV . S TRUCT URED D ESIGN TO UML D ESIGN T RANSFO RMATION T ECHNIQUES In this section, w e will analyze structure d design techniques that gene rate UML Object - oriented design based on evaluat ion crit eria discuss ed earlier . WCSIT 1 (1), 10 -16, 2011 13 A. Documentatio n Maintenance: DFD by Means o f UML. [20] In docum entation mainten ance DFD by mean s of UML, authors generate several UML diagrams from DFD. UML diagram s generated from DFD are Use-Case Diagram , Class Diagram and I nt eraction Diagram. Context diagram of DFD transform into Use-Case diagram by mapping d ata stores and sinks into actors, processes into use cases and data flows into relationsh ips between process es. For class Diagram generati on, data-stores are conside red as class es, processes attached to the data stores are considere d as functi ons and all the data elem ents of data store expresse d as attributes o f classes. For interact ion diag rams, p rocesses are use d as relationsh ip betw een generated classes. B. Meta Model App roach for Mediation [21] In metamodel approach for mediati on, authors propose formal DFD metam odel. DFD m etamodel describes DFD semantics formally. DFD instance can be created by translating semantics describe in metam odel. Authors also discuss genera tion of UML models using DFD metam odel. UML models that transform using DFD metam odel are Use - Case diagr am, Class diagram and Sequence d iag ram. Incom plete and informal mapping rules are also proposed, which are used for DFD to UML models transfo rmation. Shiroiw a proposed metamodel is a media tor between DFD and UML. metam odel als o preserves DF D hierar chy structu re. The te chnique proposed by author is gen eralized as it is based on metam odel. Case study and tool support d escri ption is not provided. Generated UML models conform ance UML semantics . Context level DFD is used for transfo rmation. Both the source and target models have medium coverage because transform ation techni que does not provide rules for every construct o f sour ce and target models. Only basic const ructs of source model are transformed into the b asic constructs of target m odel. C. Framework fo r transforming Artifacts of D FD to UML. [22] In framework for transforming ar tifacts of DFD to UML, Tran et al, propose a fram ewor k that works on DFD at three levels of abstracti on. At DFD abstraction level 1, framework translates DFD into UML Use-Case d iagr am. At DFD abstracti on level 2, framework tr anslates DFD int o UML interaction diagram . Instea d of DFD, Entity Relations hip Diagram (ERD) is used as an additional artif act for generati on of UML class diag ram structure. At abstraction level 1, processes in DFD are mapped to Use -Cases, external entities are mapped as acto rs, and data stores are mapped to classes. Data flow s mapping is remained unresolved . At ab stra ction level-2, d ata flow s variations includes externa l enti ty to process, d ata store to proc ess, process to data st ore and proce ss to external entity are transform ed into interaction diag ram . Data flow variations process to p rocess are mapped o n state transiti on diagram . At abstracti on level 3, ERD compone nts including entity, ass ociation e ntity, att ributes, and relationsh ips map to class associati on, attributes and oper ations. It is also mention ed by the authors that framew or k does not address in- depth issues of transfo rmation as complete transform ation rules a re not provi ded. The p rop osed framew ork generates valid UML models. It is clearly describ e by the author that techn ique is not scala ble. No tool and aut omation issues fo r framew ork are discussed. Technique descri bes informal transfo rmation rules and is unidirecti onal. Technique is not practical, because it is partially automatable and no case study is discussed . Every construct o f sou rce model is catered in transform ation rul es, which mak es source model coverage high. T ransf ormation rules for basic co nst ruct of target model are provided, which makes targe t model c overage Medium. D. Fun ctional and Object- oriented Views in Embedded Software Modeling . [23] In Function and Object-orien ted Views in E mbedded Softw are Mo deling, Fernande s and Liliu s describe DFD and UML diagram tr ansfo rmation. T hey propose that DFD is used in an integrate d way to refine UML models including Us e - Case and Class Diagr am. They also use DFD to detail the behavior o f a system component. Authors describe that since, DFD are more expressive to represent user requiremen t, as compare to use case d iagram it should be used to represe nts user requirem ents. DFD transform ation to UML diagram s at context level is o nly applic able. Similarly , for sequence, collabor ation and c lass diagr am DFD can also be use d. Technique proposed by au thors is partially automatable. Technique generat es valid UML models. T echniqu e uses inform al rules . Rules for very few co nstruct o f source an d target model are provided, which mak e both source and targ et model c overage low . E. Tool Support for DF D-UML Model- based Transformation [24] In the paper tool support for DFD -UML, model-bas ed transform ation, authors pr opose an approach that com bines both functional and object-orient ed models for modelin g embedded system . They also implem ented a tool for transform ation between different views. Truscan et al, propose Softw are Modeling Workbench (SMW) that g athers requirem ents, create use-case d iagr am and transf orms it into non-UML, so call Initial Object Diagram (IOD). SMW also transform s DFD’s int o Class Diagr ams. Throug h transform ation scripts, basic rules are implemented to perform transform ation. Rul es are spe cific only for I Pv6 case study . In this paper, SMW tool discu ssed by authors, run script to generate class diagram . Proposed techniqu e is automatabl e and mapped only on specific case study. We consider propos ed approach impractic al for o the rs sy stems as of specific rules for IPv6 case study. F. Sy stematic Transformation of F unctional analysis into Object- Oriented Design and Imp lementation [25] In systematic transformati on of functional analy sis into Object-Ori ented Design and implem entation, the authors proposed an enhanc ed data flow diagram called DF net, which is used to specify use cases from requirem ents. T he proposed DF net is also used in transformation to generate object- oriented design. Acco rding to authors, the transf ormation between DF net is carried in d iffer ent steps. During first step, processes in DF net dealing with data stores, data buffers and external entiti es are grouped togethe r. Similarly , processes that share the same data su ch that one process out put is the input of other process are grouped separately. Next step is to generate classes from the separated gro up. Sim ilarly, use-cases are also WCSIT 1 (1), 10 -16, 2011 14 identifie d b y using DF net. The whole process is automata ble and tool support is available for transform ation. T h e proposed approach according to authors is scalabl e and pract icable. G. A Fra mework for Transformatio n Structured Analysis and Desig n Artifact to UML [26] In the paper, framework for tr ansformation structured analysis and design artifact to UML, Fries converts D FD and Entit y Relationship Dia gram (ER D) into UM L models. UML models include use-case diagra m, seq uence diagram, sta te machine diagra m and class di agram. DFD Pr ocess in le vel - 1 DFD is mapped on use-case, e xternal entity creates an actor in use-case a nd d ata flo w creat es associatio n line in use case diagram. Similarl y, for seque nce diagra m different data flows are mapped on sequence diagram. State machi nes are generated by trans forming data flows bet ween processes as event parameters. Number of processes defines nu mber of states. ERD is used for cr eation of class diagram. The proposed framework i s partially auto matable. Author partially defines transformation rules that generate valid UML models. T ransformation is not practical, as of p artial automation, tho ugh it is ap plied on a case study b ut incomplete rules are pro vided. Tran sformation i s not scalable because transformation rules ar e defined at very abstra ct level. Transformation techniq ue is partiall y automatable as rules are inco mplete. Coverage for the source model is high as it caters ever y construct of DFD but coverage fo r target model is medium a s basic co nstructs are only catered for target model. Table 11 sho ws the compar ison of all t he tec hniques based on the evaluation criteria expl ained in section 3. V. C ONCL US IO N OF S URVEY From t he co mparison of the existing DFD - to - UM L design transformation techniques (T able 4.1), we concl ude that most of the presented approaches ar e rule -based and are incomplete; do not supports model validity a nd model generalization. T echniques in majorit y o f appr oaches define abstract rules for tra nsfor mations, which do not co ver in depth transformation and automatio n issues. Fe w of the tech niques have been applied o n case study a s a pro of of concept. Although few a uthors discuss pro cess- to -pro cess co mponent data flo w b ut none of the m prop oses appr opriate solution fo r their transformation. It is observed t hat ru les ar e defined, but at abstrac t le vel, which do es not co ver in dep th transfor mation and auto mation issues. Majority of transformation techniques are unidirectional, lacks automatio n. Few o f th e transformatio n techniques have to ol support. Some of the techniques ha ve been applied on case studies, which are not comprehensive and m ajo rity o f them are partially automated. Very fe w techniques p rovide high so urce and target m odel coverage. We have found that onl y one meta model-based tec hnique exists, which too discussed transformation at very abstract level. It is also o bserved fro m the literature revie w t hat different analysts/designers ha ve their own interpretatio n of different DFD grap hical s ymbols as DFD has i nformal s yntax. Transformation appro aches ar e d ata ce ntric, which focuses on data-store for class dia gram generatio n. Although autho rs discuss p rocess- to -process component data flow but none of them proposes co mplete solutio n for their tran sfor m atio n. ER- diagram is also used as an ad ditional artifact to generate class diagram in some of the tec hniques . The comparison i n T able 4.1 sh ows that a solution for DFD to UML class d iagram transfo rmation i s needed . A solutio n that will co ver in-depth transformation i ssues b y providi ng detailed transformation r ules. Besides, tran sformation should provide so lution for the sequence that tran sformation follo ws and cater data flow transfor m atio n between p rocesses. Transformation should follow MD A transfor mation strategy because it is the latest initiative of OMG for model transformations. B y following MDA transfor mation strategy one can use the tec hnique with the e xisting M DA and Mo del Driven Engineering (MDE ) transformation appro aches. Transfor m ation should generate UML class diagram, which is the major artifact in obj ect-oriented desig n to represent system static structure. P roposed transformation s hould provide reusable modern object -oriented desig n that will be helpful for f uture maintenance. Transforma tion strategy should be based on formal DFD metam odel, so that DFD design a m bi guity and i nconsistenc y i s not reflected in generated class diagra m. WCSIT 1 (1), 10 -16, 2011 15 T ABLE 11 A NALYSIS OF THE EXISTING DFD- TO -UM L M ODEL T RAN SFORMATION T ECHNIQUES R EFERENCES [1] Yourdon, E. (1989). Modern Structured Analy sis. Yourdon Press. Upp er Sad dle Ri ver , NJ. Ne wco mbe, P. and Doblar, R. A. (2001 December). Automated Transformation of legacy s ystems. Crosstalk, 14, 18- 22. [2] Dietrich, W., Nackman, I., Gracer, L. (1989), Saving a legacy with objects. I n Proceedings of OO PSLA, pp. 77-88. [3] Ke ndall, K. E., Kendall J. E. (1999). System Analysis and Design. (3rd E dition), Prentice Hall . [4] Larman, C. (2004). Apply ing UML and Patterns: An Introduction to Object-Oriented Analysis and Design. (3rd Edition), Addiso n Wesley Professional. [5] MDA (Model Driven Architecture), (2001 July). Retrieved on August 2006, Document or msc/2001 - 07 - 01 from http://www .omg.org/ [6] UML (Unified Modeling Language): Superstructure. Version 2.1.2. Object Management Group (OMG). Document 07- 11 - 02.pdf From ht tp://www.omg .org/spec/UML/2.1.2/ [Accessed: March 1, 2008] [7] Booch, G., Jacobson, Rumbaugh, J. I. (2005). Unified Modeling Language User Guide. (2nd Edition), Addison Wesley, Upper Saddle R iver, NJ. [8] Booch, G., R umbaug h, J., Jacobson, I. (199 9). The Unified Modeling L anguage User G uide. Addison Wesl ey. [9] Liu S., Wilde, N. (1990). I dentifying objects in a convent ional procedural lang uage: an e xample of data de sign recov ery. In: Procee dings of Conference on Software Maintenance, San Diego, CA, IEEE CS Pre ss, pp. 266-271. [10] Jacobson, I ., L indstrom, F. (1991). R e-engineering of old systems to an object -oriented archite cture. I n: Proceedings of OOPSLA , pp. 340 -350. [11] Livadas, P.E. and Johnson, T. (1994). A new ap proach to finding objects in progr ams. Journal of S oftware Main tena nce: Research and Pra ctice, vol. 6, pp. 24 9-260. [12] Gall, H. an d Klösch, R. (1995). Finding objects in procedural programs: an alternative approach. In: Proceeding of Transformati on Technique Nam e Analysis Par ameter Automatable Tool Support Additional Artifa ct Used Output Artifact Case Study Transformati on Level UML Confor mance Scalability Direction Input Model Coverage Target Model Coverage DFD Level Used Documentation maintenance: DFD by means of UML Partial No NIL Use-Case, Class & Interaction Diagram No Rule Based Yes No Unidirectional Medium Medium Context Meta-Model Approach for Mediation Yes No NIL Class Diagram. No Meta-Model Based Yes No Unidirectional Medium Medium Context Framewo rk for transforming Artifacts of DFD to UML. Partial No ER Diagram Use-Case, Class, Sequence & State Chart Diagram Yes Rule Based Yes No Unidirectional Medium High Context Functional and Object-oriented View s in Embedded Softw are Modeling. Partial No NIL Class, Use- Case & Interaction Diagram No Rule Based No No Unidirectional Low Low Context Tool Support for DFD-UML Model-based Transformation Yes Yes NIL Class Diagram Yes Rule (script) Based No No Bidirectional Medium High Extended Syste matic Transformation of Functional analysis into Object-Oriented Design and Impleme ntation Yes Yes DF Net Class Diagram Yes Rule Based Yes Y es Unidirectional Medium Medium Ex tended A F ramework for Transformation Structures Analysis and Design Artifact to UML Partial No ER Diagram Class, Use- Case, Sequence & State Chart Diagram Yes Rule Based Yes No Unidirectional Medium High Context WCSIT 1 (1), 10 -16, 2011 16 2nd Working Conference on Reverse Engineering, Toronto, Canada, I EEE CS Press, pp. 208-216. [13] Newco mbe, P. an d Kotik, G. (1995). Reengineering procedural into object -orie nted systems. I n: Proceeding of 2 nd Working Conference on Reverse Engineering, Toronto, Cana da, IEEE CS Press, p p. 237-249. [14] Subramaniam, G.V. and Bwirne, E. J. (1996). Deriving an object mo del from le gacy F ORTRAN code. In: Proceeding of International Confe rence on Software Maintenance, Mon terey, CA, IEEE CS Pre ss, pp. 3-12. [15] Cimitile, A., De Lucia, A., Di Lucca, G.A., Fasolino, A.R. (1997). Identifying objects in legacy systems. In: Proceeding Of 5th International Workshop on Program Comprehension, Dearborn, Mich igan, IEEE CS Pre ss, pp. 138-147. [16] De Lucia, G.A., Di Lucca, G.A., Fasolino, A.R., Guerra, P., Petruzzell i, S. (1997). Migrating Legacy Systems towards Object-Oriented Platforms. In: 13th International Conference on Softw are Maintenance ICSM'97. [17] De Lucia, G.A ., Fasolino, A.R., Carlini, U. D. (2000). Recover ing Class Diagrams from Data-I ntensive Legacy Syste ms. I n: 16th IEEE International Conference on Software Maintenance I CSM'00. [18] Alabiso, B. (1988). Transformation of Data Flow Analysis Models to Object-Oriented Design. In: ACM SIG PLAN Notices, 23 (Nove mber), 335-353. [19] Geor ge, J., Carter, B.D. (1996 Mar). A strategy for mappin g from fu nction oriented softw are model s to object-oriented software model s. ACM Software En gineering Notes, vol. 21, no. 2, pp. 56-63. [20] Shiroiwa, M., Tokuda, H., Miura, T. (2002). Documentation Maintenance DFD by Means of UML. In: proceeding of Information Sy stem and Enginee ring (ISE). [21] Shiroiwa, M., Miura, T., Shioy a, I. (2003 ). Meta model approach for mediation. In: Proceedings of the 27th Annual International Comp uter Software and Applications Confe rence (COMPSAC ’0 3) IEEE Pre ss, 480 -485, Ne w York. [22] Tran, T. N., Khan, K. M., Lan, Y.C. (2004). A framewor k for t ransforming arti fact s from data flow dia grams to UML. In: Procee dings of the 2004 I ASTED Int erna tional Co nference on Softw are Engineering (Innsbruck, Austria, Feb. 17-19, 2 004). ACTA Press, Calgary, AB, Canada. [23] Fernandes, J. M ., Lilius, J. (2004). Functional and object - oriented views in embedded software modeling. In: Proceedings of the I EEE Inter nati onal Confer ence and Workshop on the Engineering of Computer- Based Systems (ECBS ’04) (Brno, Czech Republi c, May 24-27, 2 004). pp 378-387, I EEE Pre ss, New Yo rk. [24] Truscan, D., Fernandes, J. M., Lilius, J. (2004). To ol supp ort for D FD-U ML model-based transformations. I n: Proceedings of the IEEE International Confere nce and Workshop on the Engineering of Computer- Based Systems (ECBS ’04) (Brno, Czech Repu blic, May 24-27, 2004). Pp 388-397, I EEE Pr ess, New Yo rk. [25] Ta n, H. B. K., Yang, Y ., Blan, L. (2006 February ). Syste matic transformation of funct ional a nalysis model in O O design and implementation. IEEE Transaction on Softw are Engineering. 32( 2), pp 111-135. [26 ]Fries, T. P. (2006). A Framework for Transforming Structured Analy sis and Design Artifacts to UML. I n: Procee dings of the 2 4th annual confere nce on Design of communication, pp 105 – 112, Myrtl e Beach, SC, USA .