EVOLUTION OF SOCIAL DEVELOPER NETWORK IN OSS: SURVEY Anil Kumar 1 1 PG Scholar, Galgotias Univrsity, Gratr Noida, UP, India Abstract Social Dvlopr Ntworks ar built of complx hirarchical social ntwork which forms rlationship information btwn softwar ntitis. Rcnt rsarch trnds suggstd many diffrnt succssful mthods on Social Dvlopr Ntwork to analyz its growth and advancmnt of ntwork structur. Aim is to prsnt brif survy on th analysis, mthod, algorithm usd in dvlopr social ntwork and tchniqus and tchnologis availabl to analyz prformanc of th dvloprs associatd ntwork. Furthr discuss th growth, xtnsion of its rsarch fild, xisting mthodologis and approach, xtndd tasks and tchniqus in dvlopr social ntwork and OSS. Kywords: Social Dvlopr Ntwork, Opn Sourc Softwar, OSS, Social Ntwork, Community Structur, dvloprs, bug rsolving procss, dfcts ----------------------------------------------------------------------***-------------------------------------------------------------------- I. INTRODUCTION 2. AN OVERVIEW OF SOCIAL DEVELOPER Social Dvlopr Ntwork (SDN) is building block of softwar ntitis ntwork and combin svral softwar projct ntitis. Social Group can b rintrprtd as SDN which is bhavior of svral nod-dg rlationships. It consists of nods for xampl Dvlopr, Projct, Fil, Author, Watchr, Usr and dg which rprsnt rlations as works, mang, commit, commnt, bug-rport tc. Ths dvlopr ntworks ar illustratd graphically to visualiz ntwork rlationship whr nods ar rprsntd having cor rol in th ntwork and dg ar rprsntd as task prformd. W prform ntwork analysis by prforming oprations on datast, xcut algorithm and showing noddg rlational activitis through visual graphical rprsntation. Whil analyzing social ntwork, Cntral nod masurmnt, Ranking pags and prioritizing masurs coms into pictur. Svral masur xists to analyz hiddn ntwork proprtis. SDN hav bn intrsting ara ovr past svral yars. Analyzing social dvlopr ntwork might point out xprts in undrstanding th social structur and assists nw comrs or activ managrs to manag thir social activitis. Our papr dals with singl point viws about prvious and currnt activitis in th Social Dvlopr Ntwork (SDN). Growth of opn accss projct movmnt startd svral yars ago. Continuous growth of softwar businss activitis in dmonstrats financial succss. Svral softwar product vndor lik Googl, RdHat ar activ organization to facilitat OSS. Catgorizing svral diffrnt platform of OSS for xampl Oprating Systm (Linux), DBMS (MySQL), Wb Contnt Managmnt (Zoomla and Wordprss), Businss Softwar tool (Magnto). NETWORK TYPES 2.1 Htrognous Social Ntwork A htrognous social ntwork holds topological information and rlational information and consists of mor ntity typs. Each vrtx is rcognizd using its nighbor rlations. A rlation is combination of dirctly and indirctly connctd nods and connctors. Htrognous Social Ntwork HSN(Vrtics, Edg, Labl) is dirctd labld graph, whr Vrtics is a limitd group of nods, Labl is a limitd st and Vrtics Labl Vrtics is a limitd group of dgs [9]. Most algorithm for xampl Mta-path analysis, ranking, similarity analysis and group similar activitis ar opration mainly prformd in this ntwork. 2.2 Homognous Social Ntwork Homognous social ntwork assums singl typ of vrtics and dg rlations. This kind of analysis mthod gnrally producs loss of information in th procss. In homognous social ntwork, vrtics dnot ntity and dg shows rlations. Availabl algorithm and mthods xampl ranking, similarity sarch, clustring and group similar activitis and association rlation prdiction 2.3 Multidimnsional Social Ntwork Dnots multipl kinds of rlationships Multi-dimnsional ntwork containing vry dimnsion constituting usr association at ach sit.g. Facbook, Twittr, YouTub and Orkut tc Li Tang t al. [13 ] Usrs rlatdnss with ach othr on svral activitis lads to multipl ntwork. Xutao Li t al. [26] stablish community rlation in multi-dimnsional ntwork by calculating th similarity btwn two itms in Volum: 03 Issu: 04 Apr-2014, Availabl @ http://www.ijrt.org 412
sam dimnsion (ntity) or diffrnt dimnsion (ntity) from th ntwork basd on probability distribution of ach dimnsion or ntity. Thir rsult shows that purposd algorithm is ffctiv and fficint. 3. LITERATURE SURVEY Social dvlopr ntwork has procdd in a dcad. In this survy papr, I intndd to show th rlatd work of crucial author's viws and thir contribution towards social dvlopr ntworks. Ovrall survy can b catgorizd in diffrnt sctions lik community structur dtction, bug localization, social dvlopr ntwork and collaboration in bug fixing procss, bug triaging. Ky Componnts Social Ntwork and Community in OSS Social Dvlopr Ntwork and Rsolving Procss Prdiction Localization Triaging 3.1 Social Ntwork and Community in OSS W Scacchi [4] studis th xpctd rlationship in social structur of dvlopmnt tam and sourc cod in opn sourc softwar. Opn sourc softwar dos not always adhr traditional softwar nginring practic but kind of complx structur of socio-tchnical procsss. Q Hong t al. [1] discovr that dvloprs and thir rlationships in social dvlopr ntwork chang continuously mad in cor dvlopr tam. Thy compard structur and volution of dvlopr as codr social ntwork with popular gnral social ntwork (GSN) such as Twittr, Facbook and Amazon. Thir study also analysd how dvlopr social ntwork grow ovr tim and ffct of DSN and growth of topological rlation of th dvlopr ntwork. Thir rsults rval strong vidnc of th community structur, maintains world charactristics ovr tim and dnot a gradual nhancmnt of community structur insid dvlopr social ntwork. Andrjs Jrmakovics t al. [11] Approach for visualizing softwar dvlopr ntwork for V C S rpositoris. Thy calculats similarity among dvlopr basd on similar fil modification. Built collaborating dvloprs ntwork and applis filtring mthods to incras th modularity of th dvlopr ntwork. Thy assign wight to all fils and comput dvloprs similarity. Zhou t al. [12] analyzd diffrnt dimnsion of softwar modification including chang significanc or sourc cod dpndncy lvls. Thy took a st of faturs from th two diffrnt sourc and proposd a Baysian ntwork for chang prdiction. Mrging clos changd ntitis and dpndncy rlation, approach can prdict undrlying uncrtainty. Th study on two mdium-sizd OSS projct indicats th fasibility and productivnss of our approach compard to prvious work. Matthw Van Antwrp and Grg Mady [10] dscribs significant considration of OSS in projcts popularity. Thir rsults show xisting dvloprdvlopr rlations ar mark of prvious and coming projct popularity. M A Baliiro t al. [14] dvlopd usr-dfind functional option calld OSS-Ntwork, which prforms study on ntwork communitis using social ntwork. OSS-Ntwork assists to dcras th difficultis of rsarchrs in cod, bug rsolving procss, and usr support. Provids an usr dfind and functional option of rtriving data, prforming obsrvation and analysis. Thung Frdian t al. [7] Thy invstigatd sourc cod construction forming ntwork in GitHub. Distinguishd lading dvloprs and projcts GitHub sub-ntwork. It hlpd dvloprs in undrstanding dvloprs and projcts association. Found out rlationship btwn projcts, dvloprs and influntial projcts. Thy analyzd that projcts ntworks in GitHub. Projct ntworks only nd common dvlopr stablishs rlation btwn projcts. Finding influntial projcts and dvlopr on th basis of pag rank givs mor dvlopr with many projcts. Thir rsults shows that distribution of projct-projct ntwork graph gnrally follows powr law whil dvlopr-dvlopr dos not. P Bhattacharya t al. [8] Visualiz volution in softwar projct. Analyzd softwar volution and build forcastr to facilitat dvlopmnt work and maintnanc. Thir rsults analysis on lvn major projcts including Eclips, Firfox, shows forcastr can dtct significant structural modification. It might assist dvlopr to masur bug svrity, prioritizing dub fforts and prdict dfcts-pron rlas. Nicolas Lopz [5] xplord topic-wis modl can assist undrstanding volution of softwar systm. H studid volution of topics taking dvloprs rlation in account. H analyzd topic pattrn of dvlopr projct works. Considrd modification that xpand OSS projcts in co-systms. His rsarch dmonstrats that dvloprs usually tak part in multipl projcts that rlats cosystm and assists dvloprs to gt xprincd working on spcific topic. Hirarchical dirctoris structur and fils of softwar program is pin point for rsarch work 3.2 Social Dvlopr Ntwork and Rsolving Procss K. Crowston and J. Howison [2] studid popl of ntwork in bug rsolving procss. Thy formd social structurd links btwn dvloprs on thir svral co-occurrnc information in bug rports in th bug chcking systm. Thir work targts 120 diffrnt projcts from SourcForg to study th cntralization problm and aims to gain communication pattrn in Opn softwar systm projcts. Rsult shows softwar tam varis in communication cntralization in th projct. Amit and Avdsh [6] studid volution and growth of social dvlopr ntwork for Eclips projct using zilla databas. Thy dmonstratd how ntwork proprty lik Avrag Path lngth, Modularity tc has impact on attributs charactrizing productivnss of bug rsolving procss. Rsult dmonstrat good rlation btwn dvlopr ntwork and Volum: 03 Issu: 04 Apr-2014, Availabl @ http://www.ijrt.org 413
proprtis distinguishing bug rsolving procss. Igor Stinmachr t al. [3] prsnts study on nw comrs joining projcts procss sinc thy ar important and potntial contributor to th projct growth and productivity. Thir study analyz th difficultis facd by nwly joind mploy and trid to vrify politnss, usfulnss of th quris or rplis rcivd by th nw comrs in th communication with tam managr. Thir rsults shows that rtntion rat is as blow as 18% in th mailing list and 13% in th issu managr and pointd out som possibl caus for laving th projct. Ovrall analysis shows nw comr ar not much intrstd to join projct du to lack of propr rspons or lack of clarification on spcific doubt. Thy trid to undrstand how dvloprs collaborat on projct and how th nwcomrs bhavior changs in communitis bcaus it hlps managmnt to tak dcision on rtntion. Yasutaka Kami t al. [16]. idntifid dvloprs rsolving bug who taks considrabl tim to drill down function or som part of sourc cod to rviw, fix and tst it vn aftr critical fils ar known. Dvloprs can rchck risky changs at tim of updating. Thy studid six opn sourc and som businss projcts from various domains basd on charactristics of softwar chang, numbr of addd lins and dvlopr xprtis. 3.3 Prdiction Zimmrman and Nagappan [17] A dpndncy graphs compard rlation btwn sourc cod. Inxprincd practitionr hav lss knowldg. Prdicting failur in post rlas lacks ffctiv allocation of rsourc and inxprinc in cod complxity. Managrs rcogniz xprt units to tackl critical dfcts. Thir work was xamind on srvr platform. Rcall rsult was 10 prcnt gratr than xpctd for complxity matrix modl ovr ntwork masurs. In th ntwork masur, dvloprs xprincd critical twic of complxity mtrics. Kim Hrzig t al. [18] All prdictd bugs ar not xpctd actual dfcts but rsults in nw faturs. rports ar classifid wrongly du to misclassification of dfcts and fatur. S. Shivaji t al. [19] dmonstrat bttr bug prdiction mthodology. Implmntd fatur idntification tchniqu in fault prdiction. Thy invstigatd svral faturs collction mthods to improv prformanc. Implmntd significanc attributd valuation filtr mthod with Bays classifir. Analyzd zilla and C V S to prdict bugs. Diffrntiatd dfcts btwn bug rpository and ach cod locations in clips projct. Figurd out pr and post rlas. calculatd total dfcts basd on ach location sinc sourc location was fixd. Exprimnt shows complx cods throw highr dfcts. Dongsun Kim t al. [20] prsntd two-phasd dfct assumption modl. Chcks whthr input bug found in xisting fil. Rsults shows 70 % of th assumption wr corrct to point bug containing fils. [Wong t al., 2007] proposd many spctra mtrics, assign lss wight to pass tst cas if particular statmnt xcutd by mor pass tst cass. Usd Simns Tst tool as bnchmark. [XiaoYuan Xi t al., 2010] proposd a post-ranking of program statmnts valuation. Thy groupd statmnts into two suspicious groups according to huristics. 3.5 Triaging A bug information is assignd, dvlopr vrify and rsolvs th bug. Potntial dvlopr is promotd with nw issu. Gaul Jong t al. [28] purposd markov chain graph modl to captur bug rassignd history. It discovrd tam hirarchy and slctd appropriat dvlopr for nw activitis. Thir modl raisd th prdiction accuracy by 23 prcnt in compar with traditional bug triaging approach. John Anvik and Gail C. Murphy [29] dmonstratd approach to gnrat rcommndrs for tim-saving dvlopmnt procss. Thir approach rsults in mor accurat whn prformd on fiv opn sourc cod projcts in bug rsolving procss. Wigin Zou t al. [30] usd fatur and instanc slction tchniqu for incrasing corrctnss of bug triag. Thy analyzd on Eclips data st. Rsults shows nw and small training st can gnrat bttr accuracy than th xisting original on. 4. VISUAL ILLUSTRATION OF SOCIAL DEVELOPER NETWORK 3.4 Localization Th basic concpt bhind th Localization is to calculat statistics charactrizing a softwar program's runtim bhavior ovr multipl xcutions. Erronous sourc cod contains critical bug which rsults in failur at runtim. It causs incrasd in rsourc cost and dcrasd productivity. localization assists program managr and practitionr to idntify sourc cod and rsolv bug in minimizd tim to incras productivity. Localizing or locating dfcts in th cod hav bn challnging task. Zimmrman t al. [15] Fig 1 Social Dvlopr Ntwork Volum: 03 Issu: 04 Apr-2014, Availabl @ http://www.ijrt.org 414
Abov dirctd ntwork graph contains fiv distinct typs of labld nod, dgs and rlationships. This rprsntation uss dvlopr ntwork to show social rlationship among svral distinct nod. Visual rprsntation can b undrstood from Top nod as Dvlopr nod followd by Projct nod, nod and Commnt nod rspctivly. Rlationships btwn nods as Dvlopr->Projct->->Commnt and Dvlopr->Projct->Languag. Dvlopr to Projct (labld as D), Projct to Languag (labld as L), Projct to (labld as B) and Projct to Commnt (labld as C). Dpictd social ntwork diagram dmonstrats that Dvlopr Id (8) work on Projct (mmcachd) and (Folly), writ cod in languag (C++), Ids (2312, 40867) and (15637, 15746, 15720) rspctivly having svral commnts in ach projcts. 5. SUMMARY SOCIAL DEVELOPER NETWORK TECHNIQUES AND PURPOSED APPROACH Tabl 1: Summary of survyd information of svral author's tasks, tchniqus. Charactris tics Data Sourc Mthod Activitis Approac h Rports Information Rport Rport Sourc Cod Fil and Function Variabl Sourc cod and Rport Rport Rport zil la datas t Fiv OSS Projc ts zil la Mozill a, Apach Sourc cod and CVS zil la and CVS Mozill a and Eclips Eclips Classificat ion Suprvis d Larning Machin Larning Complxit y Mtrics Classificat ion Mtadata analysis tracking systm Markov chain graph modl Fatur, instanc slction Triaging Triaging Prdict Postrlas failur prdictio n Chang Prdictio n Localizati on Triag Triag Anvik John Anvik and Gail C. Murphy [29] Zimmr man and Nagappa n [17] Kim Hrzig t al. [18] Hassan and Holt Zimmr man t al. [15] Gaul Jong t al. [28] Wigin Zou t al. [30] 6. CONCLUSIONS Social Dvlopr Ntwork and OSS ar compatibl and distributd. OSS is distributd softwar nvironmnt. OSS facilitat rducd cost, good prformanc, bttr quality, rducd tim, incrasd scurity in critical oprations and projct managmnt conomically and financially. Social Dvlopr Ntwork community structur built to mak rlationship btwn svral opn sourc projct. This survy dmonstratd diffrnt dvlopr ntworks, dvlopr rlationship and associatd rlationship with bug rsolving procss, localization of bug and prdicting bug in social dvlopr ntwork. REFERENCES [1] Qiaona Hong, Sunghun Kim, S.C. Chung, "Christian Bird, Undrstanding a Dvlopr Social Ntwork and its Evolution", Procding ICSM '11 Procdings of th 2011 27th IEEE Intrnational Confrnc on Softwar Maintnanc, Pags 323-332 [2] K. Crowston and J. Howison, "Th social structur of fr and opn sourc softwar dvlopmnt," First Monday, vol. 10, no. 2, 2005. [3] Igor Stinmachr, Igor Wis, Ana Paula Chavs, Marco Aurélio Grosa, "Why do nwcomrs abandon opn sourc softwar projcts?",cooprativ and Human Aspcts of Softwar Enginring (CHASE), 2013 6th Intrnational Workshop on IEEE, 2013, pag. 25-32 [4] Walt Scacchi, "Softwar Dvlopmnt Practics in Opn Softwar Dvlopmnt Communitis: A Comparativ Cas Study" In Procdings of th First Workshop on Opn Sourc Softwar Enginring, 2001, [5] Nicolas Lopz, "Using topic modls to undrstand th volution of a softwar cosystm", ESEC/FSE 2013 Procdings of th 2013 9th Joint Mting on Foundations of Softwar Enginring, Pags 723-726 [6] Amit Kumar, Avdsh Gupta, "Evolution of Dvlopr Social Ntwork and Its Impact on Fixing Procss", ISEC 13, Fb 2013, 21-23 [7] Thung, Frdian; LO, David; and JIANG, Lingxiao, "Ntwork Structur of Social Coding in GitHub" (2013). Rsarch Collction School of Information Systms (Opn Accss). [8] Pamla Bhattacharya, Marios Iliofotou, Iulian Namtiu, Michalis Faloutsos, "Graph-Basd Analysis and Prdiction for Softwar Evolution", Procding ICSE '12 Procdings of th 34th Intrnational Confrnc on Softwar Enginring, Pags 419-429 [9] Chng-T Li and Shou-D Lin, "Cntrality Analysis, Rol-basd Clustring, and Egocntric Abstraction for Htrognous Social Ntworks" 2012 Intrnational Confrnc on Social Computing ((PASSAT, SocialCom), IEEE, Spt 2012 Volum: 03 Issu: 04 Apr-2014, Availabl @ http://www.ijrt.org 415
[10] Matthw Van Antwrp, Grg Mady, "Th Importanc of Social Ntwork Structur in th OpnSourc Softwar Dvlopr Community", Procdings of th 43rd Hawaii Intrnational Confrnc on Systm Scincs - 2010 [11] Andrjs Jrmakovics, Albrto Sillitti, Giancarlo Succi, "Mining and visualizing dvlopr ntworks from vrsion control systms", Procding CHASE '11 Procdings of th 4th Intrnational Workshop on Cooprativ and Human Aspcts of Softwar Enginring,Pags 24-31 [12] Zhou Y,W urschm, Gigr E, Gall H, L u J., A baysian ntwork basd approach for chang coupling prdiction, In Procdings of th Working Confrnc on Rvrs Enginring (WCRE), IEEE Computr Socity: Washington, DC, USA, 2008; [13] Li Tang, Xufi Wang, and Huan Liu. Community Dtction via Htrognous Intraction Analysis. Knowldg Discovry and Data Mining (DMKD), 2012 [14] Marco A. Baliiro, Samul F. d Sousa Júnior, and Clidson R. B. d Souza, "Facilitating Social Ntwork Studis of FLOSS using th OSSNtwork Environmnt", Opn Sourc Dvlopmnt, Communitis and Quality IFIP Th Intrnational Fdration for Information Procssing Volum 275, 2008, pp 343-350 [15] Zimmrmann, T. Prmraj, R., Zllr, A, "Prdicting dfcts for Eclips", In Proc. 3rd Intrnational Workshop on Prdictor Modls in Softwar Enginring (Minnapolis, MN, USA, May, 2007), PROMISE 07. [16] Yasutaka Kami, Emad Shihab, Bram Adams, Ahmd E. Hassan, Audris Mockus, Anand Sinha, and Naoyasu Ubayashi, "A Larg-Scal Empirical Study of Just-in- Tim Quality Assuranc" IEEE Transactions On Softwar Enginring, Vol. 39, No. 6, Jun 2013 [17] Zimmrmann, T. and Nagappan, N., "Prdicting Dfcts using Ntwork Analysis on Dpndncy Graphs," in 29th Intrnational Confrnc on Softwar Enginring, 2007 [18] Kim Hrzig, Sascha Just, Andras Zllr, "It's not a bug, it's a fatur: how misclassification impacts bug prdiction", Procding ICSE '13 Procdings of th 2013 Intrnational Confrnc on Softwar Enginring, Pags 392-401 [19] Shivkumar Shivaji, E. Jams Whithad, Jr., Ram Aklla, Sunghun Kim, "Rducing Faturs to Improv Cod Chang Basd Prdiction", Journal IEEE Transactions on Softwar Enginring archiv Volum 39 Issu 4, April 2013, Pags 552-569 [20] Dongsun Kim, Yida Tao, Sunghun Kim, Andras Zllr, "Whr Should W Fix This? A Two- Phas Rcommndation Modl", IEEE Transaction on Softwar Enginring, VOL. 39, NO. 11, NOVEMBER 2013 [21] Mosr, R., Abrahamsson, P., Pdrycz, W., Sillitti, A., Succi,G., A cas study on th impact of rfactoring on quality and productivity in an agil tam, In Proc. of th 2nd IFIP Cntral and East Europan Confrnc on Softwar Enginring Tchniqus CEE-SET 2007, Poznan, Poland (2007). [22] Huzfa Kagdi, Michal L. Collard and Jonathan I. Maltic1, A survy and taxonomy of approachs for mining softwar rpositoris in th contxt of softwar volution, Journal of Softwar maintnanc and volution : Rsarch and Practic J. Softw. Maint. Evol.: Rs. Pract. 2007; 19:77 131 [23] Xiao Yuan Xi and Tsong Yuh Chn and BaoWn Xu "Isolating Suspiciousnss from Spctrum-Basd Fault Localization Tchniqus", In Procdings of th 2010 Intrnational Confrnc on Quality Softwar (QSIC), 2010, pag 385-392, IEEE [24] Wong, W.E. and Shi, Y. and Qi, Y. and Goldn, R. "Using an RBF nural ntwork to Locat Program s", In Procdings of th 19th Intrnational Symposium on Softwar Rliability Enginring, 2008, pag 27-36, 2008, IEEE/ACM. [25] Santlics, R. and Jons, J.A. and Yu, Y. and Harrold, M.J. "Lightwight fault-localization using multipl covrag typs ", In Procdings of th 31st Intrnational Confrnc on Softwar Enginring, 2009, pag 56-66, 2009, IEEE. [26] Xutao Li, Ng, M.K., Yunming Y, " MultiComm: finding community structur in multi-dimnsional ntworks", Knowldg and Data Enginring, IEEE Transactions on Volum:26,Issu: 4, April 2014, pp. 929-941, IEEE [27] Huzfa Kagdi, Michal L. Collard and Jonathan I. Maltic, "A survy and taxonomy of approachs for mining softwar rpositoris in th contxt of softwar volution", Journal of Softwar maintnanc and volution : Rsarch and Practic J. Softw. Maint. Evol.: Rs. Pract. 2007; 19:77 131 [28] Gaul Jong, Sunghun Kim, Thomas Zimmrmann, "Improving Triag with Tossing Graphs", Procding ESEC/FSE '09 Procdings of th th 7th joint mting of th Europan softwar nginring confrnc and th ACM SIGSOFT symposium on Th foundations of softwar nginring, Pags 111-120 [29] John Anvik, Gail C. Murphy, "Rducing th ffort of bug rport triag: Rcommndrs for dvlopmntorintd dcisions", Journal ACM Transactions on Softwar Enginring and Mthodology (TOSEM) TOSEM Hompag archiv Volum 20 Issu 3, August 2011, Articl No. 10 [30] Wiqin Zou, Yan Hu,Jifng Xuan,H Jiang, "Towards Training St Rduction for Triag",Computr Softwar and Applications Confrnc (COMPSAC), 2011 IEEE 35th Annual, Pags 576-581 Volum: 03 Issu: 04 Apr-2014, Availabl @ http://www.ijrt.org 416