Mining Project-Oriented Business Processes

ePubWU Ins i u ional Reposi o y Saimi Bala and C is ina Cabanillas Macias and And eas Sol i and Jan Mendling and Axel Polle es Mining P ojec - O ien ed Business P ocesses Book Sec ion (D a ) (Re e eed) O iginal Ci a ion: Bala, Saimi and Cabanillas Macias, C is ina and Sol i, And eas and Mendling, Jan and Polle es, Axel ORCID: h ps://o cid.o g/0000-0001-5670-1146 (2015) Mining P ojec - O ien ed Business P ocesses. In: Business P ocess Managemen . BPM 2016. Lec u e No es in Compu e Science, ol 9253. Sp inge , Cham. pp. 425-440. ISBN 1611-3349 This e sion is a ailable a : h ps://epub.wu.ac.a /7203/ A ailable in ePubWU: Oc obe 2019 ePubWU, he ins i u ional eposi o y o he WU Vienna Uni e si y o Economics and Business, is p o ided by he Uni e si y Lib a y and he IT-Se ices. The aim is o enable open access o he schola ly ou pu o he WU. This documen is an ea ly e sion ci cula ed as wo k in p og ess. The e a e mino di e ences be ween his and he publishe e sion which could howe e a ec a ci a ion. h p://epub.wu.ac.a / Mining P ojec -O ien ed Business P ocesses? Saimi Bala, C is ina Cabanillas, Jan Mendling, And eas Rogge-Sol i, and Axel Polle es Vienna Uni e si y o Economics and Business, Aus ia { i s name.las name}@wu.ac.a Abs ac . La ge enginee ing p ocesses need o be moni o ed in de ail ega ding when wha was done in o de o p o e compliance wi h ules and egula ions. A ypical p oblem o hese p ocesses is he lack o con- ol ha a cen al p ocess engine p o ides, such ha i is di icul o ack he ac ual cou se o wo k e en i da a is s o ed in e sion con ol sys ems (VCS). In his pape , we add ess his p oblem by de ining a mining echnique ha helps o gene a e models ha isualize he wo k his o y as GANTT cha s. To his end, we o mally de ine he no ion o a p ojec -o ien ed business p ocess and a co esponding mining algo i hm. Ou e alua ion based on a p o o ypical implemen a ion demons a es he bene i s in compa ison o exis ing p ocess mining app oaches o his speci ic class o p ocesses. Keywo ds: p ocess mining, p ojec s, p ojec mining, e sion con ol sys ems 1 In oduc ion Business p ocess managemen plays an impo an ole o imp o ing he pe o - mance and compliance o a ious ypes o p ocesses. In p ac ice, many p ocesses a e execu ed wi h clea guidelines and egula o y ules, bu wi hou an explici cen alized con ol imposed by a p ocess engine. In pa icula , i is o en im- po an o exac ly know when which wo k was done. This is, o ins ance, he case o complex enginee ing p ocesses in which di e en pa ies a e in ol ed. We e e o his class o p ocesses as p ojec -o ien ed business p ocesses. Such p ojec -o ien ed business p ocesses a e di icul o con ol due o he lack o a cen alized p ocess engine. Howe e , he e a e a ious uns uc u ed pieces o in o ma ion a ailable o analyze and moni o hei p og ess. One ype o da a ha a e o en a ailable hese p ocesses is e en da a om e sion con ol sys ems (VCS). While p ocess mining echniques p o ide a use ul pe spec i e on how such e en da a can be analyzed, hey do no p oduce ou pu ha is eadily o ganized acco ding o he p ojec o ien a ion o hese p ocesses. ?This wo k has been unded by he Aus ian Resea ch P omo ion Agency (FFG) unde g an 845638 (SHAPE) and he Eu opean Union’s Se en h F amewo k P o- g amme unde g an 612052 (SERAMIS). 2 S. Bala e al. In his pape , we de ine o mal concep s o cap u ing p ojec -o ien ed p o- cesses. These concep s p o ide he ounda ion o us o de elop an au oma ic disco e y echnique which we e e o as p ojec mining. The ou pu o ou p ojec mining algo i hm is o ganized acco ding o he speci ic s uc u e yp- ically encoun e ed in p ojec -o ien ed business p ocesses. Wi h his wo k, we ex end he ield o p ocess mining owa ds he co e age o his speci ic ype o business p ocess. The pape is s uc u ed as ollows. Sec ion 2 desc ibes he esea ch p ob- lem and summa izes insigh s om p io esea ch upon which ou p ojec mining app oach is buil . Sec ion 3 de ines he p elimina ies o ou wo k and p esen s an algo i hm o mine p ojec -o ien ed business p ocesses. Sec ion 4 desc ibes he implemen a ion o his algo i hm and discusses he esul s om i s applica- ion o VCS logs om a eal-wo ld enginee ing p ojec . Sec ion 5 highligh s he implica ions o his wo k be o e Sec ion 6 concludes. 2 Backg ound He e, we desc ibe he add essed p oblem and ela ed wo k. 2.1 P oblem Desc ip ion The class o p ocesses ha we discuss in his pape a e long- e m enginee ing p ojec s. These p ocesses ha e speci ic equi emen s o moni o ing. Fi s , hey a e execu ed only once acco ding o he speci ic needs o a pa icula p ojec , and only pa ially acco ding o ecu ing p ocess desc ip ions. Second, hey in ol e a ious ac o s ha ypically documen hei wo k in a semi-s uc u ed way using ex and ables. Thi d, wo k in he p ojec is usually subjec o cons ain s ega ding he s a and end and he empo al o de . Fou h, he e is ypically no p ocess engine con olling he execu ion. Fi h, e en hough hese limi a ions in e ms o aceabili y exis , he e a e usually s ong equi emen s in e ms o acking when which wo k was conduc ed. In line wi h hese obse a ions, a p ojec -o ien ed business p ocess can be de ined as an ad-hoc plan ha speci ies he asks o be pe o med wi hin a limi ed pe iod o ime and wi h a limi ed se o esou ces o achie ing a speci ic goal. Unlike epe i i e business p ocesses o which no a ions such as BPMN [12] o EPC [1] a e commonly used, p ojec -o ien ed business p ocesses may be p ope ly ep esen ed wi h PERT o GANTT models. The concep is illus a ed in Fig. 1. Documen a ion is equi ed no only explici ly as pa o some ac i i ies bu also o comply wi h no ms and egula ions ha may equi e some e idence o he ac ions being pe o med in he o ganiza ion. Documen s a e usually ee o o ma o con ain ables, a bes . The uns uc u edness o da a makes i di icul o moni o p ocesses and check ules on hem. A s a ing poin o analysis o p ojec -o ien ed p ocesses can be da a logs ha a e s o ed in So wa e Con igu a ion Managemen (SCM) sys ems ha help acking he e olu ion o da a and es o e in o ma ion i needed [19]. Howe e , hund eds o e sions o Mining P ojec -O ien ed Business P ocesses 3 P1 RESOURCES P1P2 P2 1 2 DATA PROJECT ACTIVITY 1 ACTIVITY 2 {P1,P2} {P2} ... n ... ACTIVITY n {Pn} PROJECT MINING Fig. 1. P oblem illus a ion housands o iles a e common in a single p ojec [20], which makes i imp ac ical o b owse his da a manually. Le us see an example inspi ed by a eal scena io o a p ocess o w i e a p ojec p oposal ha uses a Ve sion Con ol Sys em (VCS) o s o e he da a. The p ojec his o y, and hence, he da a p oduced, s a s when people begin o wo k on he p oposal, which in ol es a desc ip ion o he p ojec goals and mile- s ones, a di ision o asks in o wo k packages, an es ima ion o cos and esou ces equi ed, e ce e a. This in o ma ion is sp ead in he eposi o y o e se e al old- e s con aining di e en documen s, which a e la e me ged in o a single ile. I he p oposal is accep ed, he i s s ep is o o ganize a kicko mee ing and assign speci ic esou ces o he wo k packages. A hie a chical se o olde s is hen c e- a ed in he eposi o y in o de o s o e he in o ma ion gene a ed o each wo k package. As he p ojec e ol es o e ime, esou ces con ibu e by adding, emo - ing o modi ying in o ma ion o he VCS eposi o y. P ojec e olu ion is guided by speci ic no ms ha impose he execu ion o p ede ined s eps. Fo ins ance, he Eu opean no m EN5016 equi es a p elimina y Reliabili y, A ailabili y and Main ainabili y (RAM) analysis o suppo a ge s. Table 1 depic s an exce p o he log da a gene a ed, whe e he i s column (on he le hand side) indica es he commi iden i ie , he second column indi- ca es he pe son who commi ed changes, he hi d column indica es he commi da e, and he ou h column indica es he iles a ec ed and he ype o ac ion pe o med among added (A), modi ied (M) and dele ed (D). Fo he sake o simplici y, he able shows he log da a o a speci ic ime pe iod and he ac ions ela ed o a speci ic ask, namely, De ine example. Tha ask was assigned o esou ce X and was supe ised by esou ce Y and, la e on, also by esou ce Z. Exis ing amewo ks, such as Sub e sion o Gi , allow o access hei logs in di e en ways. Howe e , he co e ed in o ma ion is limi ed o ( oughly) ha 4 S. Bala e al. Table 1. Exce p om VCS log da a o he e e enced ime pe iod CID Resou ce Da e Lis o changes 1 Y 2014-11-12 11:57:46 A /example A /example/SHAPE/ToyS a ionExample.docx . . . . . . . . . . . . 3 X 2014-11-14 16:34:07 M /example/ToyS a ion.bpmn M /example/ToyS a ion.png 4 W 2014-12-15 13:49:11 D /example/Download 5 W 2015-01-08 16:06:41 A /example/Download2 6 X 2015-01-13 11:47:09 M /example/ToyS a ion 0Loop.bpmn M /example/ToyS a ion nLoop.bpmn 7 Z 2015-01-16 16:50:29 A /example/ToyS a ion 0Loop.pd A /example/ToyS a ion- eedbackZ.pd depic ed in Table 1. Especially o big p ojec s ha a e equen ly upda ed o e a la ge pe iod o ime, hese logs a e complex o analyze. The e o e, he p oblem o add ess is how o analyze and isualize he in o ma ion p oduced in p ojec -o ien ed business p ocesses such ha i can be ep esen ed in an unde s andable and manageable way by p ojec expe s and enable, a.o., he au oma ion o mechanisms o compliance checking. The ollowing p ope ies o p ojec -o ien ed p ocess logs mus be aken in o accoun o achie e his goal: (i) VCS eposi o ies consis o a hie a chy o olde s and iles which a e logically o ganized such ha wo k is g ouped in a speci ic way; (ii) p ocess ac i i ies a e no egis e ed in VCS log en ies. The e o e, such in o ma ion mus be in e ed by easoning on he eposi o y s uc u e and/o he con en o he log en ies; (iii) he g anula i y o he e en s is unknown a p io i and i needs o be de ined be o e analyzing he da a. 2.2 Rela ed Wo k The p oblem desc ibed has been add essed in he li e a u e om di e en pe - spec i es. The i s ca ego y o ela ed wo k ackles he p oblem by ans o ming i in o a p ocess mining p oblem. Consequen ly, app oaches ha e been de eloped o p ep ocess VCS da a such ha p ocess mining echniques can be applied, and hence, a business p ocess can be de i ed om he log da a. In his g oup, Kindle e al. [9,10] de eloped an algo i hm o ex ac ing so wa e p ocesses ha a e mapped o Pe i Ne s. Ac i i ies, which a e no explici in he logs, a e disco e ed om hei inpu and ou pu a i ac s. Howe e , s ong assump ions a e made on he ilenames as well as on he so wa e p ocess li ecycle. Rubin e al. in [15] ad- d essed he p oblem o enginee ing p ocesses ha a e no well documen ed and a e usually uns uc u ed. They p o ided a b idge om Kindle e al.’s app oach o P oM [5] in o de o mine di e en p ocess pe spec i es, such as pe o mance social ne wo k analyses. Rubin e al. [16] applied p ocess mining o he ou is ic Mining P ojec -O ien ed Business P ocesses 5 indus y and ob ained use p ocesses om web clien logs pu suing he goal o imp o ing he so wa e sys em by analyzing he unde lying p ocess. Poncin e al. [14] de eloped he FRASR amewo k o p ep ocessing so wa e eposi o ies o ans o m he VCS da a o logs ha con o m o he p ocess mining e en log me a model [4] as u ilized in P oM [5]. Howe e , hese app oaches dis ega d he single-ins ance na u e o p ojec -o ien ed business p ocesses and ea hem as p ocedu es ha can be epea ed o e ime. The second ca ego y o ela ed wo k ocuses on he isualiza ion o VCS da a o di e en pu poses. Se e al app oaches s udy he in e ac ion among de elop- e s o e ime om a isualiza ion poin o iew. Fo ins ance, Ogawa and Ma [11] d ew s o yline pa hways o show he s o y o each de elope ’s con ibu ion. O he app oaches analyze and isualize VCS da a a ile le el in o de o disco e ile e sion e olu ion. Voinea and Telea [20] in oduced an in e ac i e na iga ion me hod o su ile e sion e olu ion as well as wo me hods o clus e e sions o he same ile in an abs ac ion laye . Wu e al. [22] also isualized he e olu- ions o en i e p ojec s a ile le el, emphasizing he e olu ion momen s. Finally, se e al app oaches s udy change p edic ion wi h he aim o disco e ing p edic- ion pa e ns ha can help in he p ocess o so wa e de elopmen [24,23]. The app oaches men ioned in his ca ego y as well as o he s ha apply simila ech- niques [6,8,3] ocus on s udying so wa e e olu ion om di e en s andpoin s. Howe e , he goal pu sued di e s in all cases om ou goal in ha hey a e no in e es ed in disco e ing p ojec s asks ou o he log da a, and hence, hey lack an explici no ion o wo k s uc u e ha we need o conside o ou pu pose. Ou app oach combines ideas om bo h a eas, as we aim a iden i ying asks like in he app oaches ha ely on p ocess mining, bu we mus clus e he da a in an app op ia e way, o which echniques de eloped in he app oaches ha pu sue isualiza ion may be adap ed o ex ended. 3 Mining VCS E en Da a He e, we i s o malize he no ions encoun e ed in he p ojec mining se ing. Then we de elop an app oach o acqui e a hie a chical o e iew on he p ojec om a eposi o y pe spec i e. 3.1 P elimina ies Ve sion con ol sys ems (VCSs) a e used in p ojec s o ensu e eliable collabo- a ion. We build ou app oach on VCS. Typically, he wo k low in VCS is ha people wo k on iles (e.g., ex , sou ce code, sp ead shee s) and commi hem o he cen al eposi o y. P ojec pa icipan s commen on hei commi s so ha o he pa icipan s can be e unde s and he na u e o he changes o he iles. Le Fbe he uni e se o iles. Files a e o ganized in a ile ee. The e o e, each ile ∈Fhas one pa en ile. The only ile wi hou a pa en ile is he oo ile. We cap u e his in o ma ion in he pa en ela ion Pa en :F×F. Fo example, le p∈Fbe he pa en o ile c∈F, hen ( p, c)∈Pa en . The 6 S. Bala e al. ansi i e closu e on he pa en iles is gi en by he unc ion ances o :F→2F ha e u ns he se o iles along he pa h o he oo . When p ojec membe s did a ce ain amoun o wo k and wan o sa e hei cu en p og ess, hey commi he changes o he VCS. We de ine changes on iles as he e en s o in e es on he lowes g anula i y. De ini ion 1 (E en ). Le E be he se o e en s. An e en e ∈E is a ou - uple ( , o, s,k), whe e – ∈Fis he a ec ed ile o he e en . –o∈O={added, modi ied, dele ed}is he change ope a ion on he ile wi h ob ious meaning. – s ∈TS =N0 ep esen s a unix ime s amp ma king he ime o he e en occu ence. –k∈Σ∗is a commen in na u al language ex . Fo e en s e= ( , o, s,k) we o e load , o, s, and k o be used as accesso unc ions. Fo example, is he unc ion :E→Fmapping an e en o i s a ec ed ile. P ojec pa icipan s can commi a numbe o changes o di e en iles a one s ep. The e o e, we de ine he no ion o commi s as ollows. De ini ion 2 (Commi ). A commi Cis a se o e en s sha ing he same ime s amp and commen , i.e., ∀e,e0∈ C : s(e) = s(e0)∧k(e) = k(e0). Addi ionally, each e en in a commi a ec s di e en iles, i.e., ∀e,e0∈ C :e6=e0→ (e)6= (e0). Usually, i is in he hands o p ojec pa icipan s, when hey decide o commi changes o he VCS. In he ex eme case, he e could be only a single commi made in a p ojec ha adds all iles o he eposi o y. No e ha his ex eme p ac ice would ende he use o a VCS obsole e. On he con a y, i is common p ac ice o egula ly pe o m commi s in o de o secu ely s o e wo k p og ess and o educe he chance o con lic s [13,7]. Con lic s occu , when ano he pa - icipan commi ed changes o a ile ha is being commi ed and can cause ex a wo k. Based on hese insigh s, we make he assump ion ha commi s a e egula ly made du ing wo k. P ojec s a e decomposed in o wo k packages. We assume a hie a chical wo k package s uc u e o a p ojec , such ha a wo k package can ha e sub wo k packages. Fu he , he amoun o wo k in a single wo k package need no be done in one single ime span, bu i can be spli in o se e al ac i i ies. Ac i i ies ha e a s a and end ime, and subsequen ac i i ies can ha e idle pe iods in be ween. Thus, we de ine p ojec s as ollows. De ini ion 3 (P ojec ). A p ojec Pis a uple (W, S, A, α, ω, β), whe e –Wis he se o wo k packages in he p ojec . –S⊆W×Wis he ela ion ha hie a chically decomposes wo k packages in o a ee s uc u e. –A is he se o ac i i ies ha a e conduc ed in he wo k packages. Mining P ojec -O ien ed Business P ocesses 7 –α:A→TS is he unc ion ha assigns a s a ime o ac i i ies. Ac i i ies a e o de ed by hei s a imes. –ω:A→TS is he unc ion ha assigns an end ime o ac i i ies. –β:A→Wis he mapping unc ion ha maps ac i i ies o hei co espond- ing wo k packages. No e ha his de ini ion e lec s an ac i i y cen ic iew on p ojec s. The de ini ion delibe a ely omi s u he dimensions, e.g., cos s, esou ces, isks. The idea is no o cap u e p ojec s in e e y de ail, bu o ocus on he wo k packages o a p ojec o ob ain an o e iew o he wo k ha is being done. We a e in e - es ed in when wo k has been s a ed in a wo k package, and when wo k packages ha e been done. This in o ma ion can be de i ed om he ac i i ies associa ed o he wo kpackages. An ob ious assump ion is ha he wo k package s a s wi h i s i s ac i i y, and ends when i s las ac i i y is comple ed. Based on hese no ions, we can de ine he ask o p ojec disco e y as econ- s uc ing he p ojec P om a se o low le el e en da a E. In he ollowing, we p esen an app oach o his p oblem. 3.2 P ojec Disco e y Technique Fo p ojec disco e y om he VCS commi his o y, we need o iden i y ac i i ies ha a e pe o med, associa e he ac i i ies o wo k packages and ec ea e he wo k package s uc u e o he p ojec . Ou aim is o c ea e a hie a chical model ha p o ides an o e iew o he p ojec wo k. The e o e, we ha e o iden i y he s a and end imes o ac i i ies and o wo k packages be o e we can isualize he p ojec wo k. The inpu o he echnique is he log ha is s o ed in he VCS. The challenge is ha he aw log only eco ds commi s on he ile sys em le el and in o ma ion on ac i i y le el is missing. Howe e , we can deduce ac i i y in o ma ion om e en s based on he ollowing assump ions. A1: Meaning ul ile ee s uc u e. The ile ee s uc u e in a p ojec ep- esen s i s wo k package s uc u e. Tha is, he knowledge wo ke s o ganize hei wo k in a ile hie a chy ha e lec s he p ojec s uc u e. A2: Local changes. Ac i i ies in a wo k package a ec only iles o he wo k package olde , o in he co esponding sub- ee in he ile ee s uc u e. A3: F equen commi s. Commi s o he VCS a e egula ly pe o med, when conduc ing wo k in an ac i i y. disco e y s a ed P ep ocess log in o a se o e en s VCS log e en s E Agg ega e e en s o ac i i ies ac i i ies A Iden i y wo k packages o ac i i ies wo k packages W Agg ega e wo k packages wo k package s uc u e S Compu e wo k package cha ac e is ics p ojec P Fig. 2. P ojec disco e y echnique o e iew as BPMN p ocess model. 8 S. Bala e al. No e ha assump ion A1 can be seen as a s ong assump ion on he ile ee s uc u e. Ne e heless, we a gue ha e en i A1 is no en i ely me , he agg ega ion o wo k in o ma ion on he ile ee hie a chy p o ides a aluable iew on he p ojec . Figu e 2 shows he di e en s eps o he echnique. We desc ibe each o hem in de ail. S ep 1: P ep ocessing. The i s s ep is o ans o m aw logs o e sion con ol sys ems (which migh be g ouped by commi s) in o a lis o e en s as speci ied in De ini ion 1. This s ep is easily done by eplica ing he in o ma ion on commi le el o be con ained in he e en s. The ou pu is a se o e en s E. S ep 2: Agg ega ing e en s o ac i i ies. Gi en he se o e en s E ha we ga he ed om a e sion con ol sys em, he nex s ep is o iden i y he ac i i ies o which he e en s belong. No e ha we do no know he ac i i ies o he p ojec in ad ance, bu need o in e hem based on he e en s. Each e en a ec s a single ile in he ile hie a chy. ime c1c2c4 c2 c3 ? a3 obse ed ac i e ime a3' adjus ed ac i e ime c3 c4 c c1 Fig. 3. Adjus men o ac i i y s a ime α. Based on assump ion A2, we a e in e es ed in ac i i ies conduc ed in a wo k package, ha is, we il e o he e en s ha a e con ained in he gi en ile o i s child en. Fo e e y ile o in e es , we selec he se o e en s a ec ing he ile o i s child en as E ={e∈E| = (e)∨ ∈ances o ( (e))}. The ask is hen o ind he ac i i ies which emi ed he se o e en s E . We ely on assump ion A3, which s a es ha du ing an ac i i y, we expec mul iple commi s. Assump ion A3 allows us o conclude ha i we do no obse e commi s o a longe pe iod o ime, he e is no ac i i y being pe o med in he wo k package. To his end, we adop he abs ac ion echnique by Baie e al. [2] and allow he domain expe o o mula e ules o agg ega ing e en s o ac i i ies based on bounda y condi ions. Assuming ha people equen ly commi hei p og ess (A3), we can speci y a bounda y condi ion based on he empo al dis ance o p e ious e en s. Fo example, we can speci y ha a ime pe iod o se en days wi hou a commi is a bounda y condi ion. As he esul , we ob ain he mapping om e en s o hese ac i i ies, which we call γ :E →A in he emainde o he pape . The se o disco e ed ac i i ies iden i ied o he wo k package based on gi en bounda y condi ions is hen A ={a|e∈E , γ (e) = a}. We also Mining P ojec -O ien ed Business P ocesses 15 Fig. 6. Do ed cha om P oM Fig. 7. Cha om Disco plo ing he e en s o e ime. 6 Conclusion In his pape we add essed he p oblem o mining and isualizing p ojec - o ien ed business p ocesses in a way ha is in o ma i e o manage s. We de- ine an app oach ha akes VCS logs as inpu o gene a e Gan cha s. Ou algo i hm wo ks unde he assump ions ha eposi o ies e lec he hie a chical s uc u e o he p ojec , each wo k package is con ained in a co esponding di- ec o y and p ojec membe s commi hei wo k egula ly du ing ac i e wo king imes. The app oach was implemen ed as a p o o ype and e alua ed based on eal-wo ld da a om open sou ce p ojec s. In u u e wo k, we aim o ex ac u he de ails o he VCS logs in o de o calcula e me ics ha app oxima e he wo k e o . We plan o in es iga e on how he p ojec mining app oach is a ec ed by p ojec cha ac e is ics. Fu he mo e, we wan o u ilize s a is ical me hods o be e es ima e he bounda ies o he ac i i ies and wo k packages. Finally, we ha e al eady inco po a ed eedback om manage s and plan o ex end hese o ull use s udies. Re e ences 1. an de Aals , W.: Fo maliza ion and e i ica ion o e en -d i en p ocess chains. In o ma ion and So wa e Technology 41(10), 639–650 (1999) 2. Baie , T., Mendling, J., Weske, M.: B idging abs ac ion laye s in p ocess mining. In o ma ion Sys ems 46, 123–139 (2014) 16 S. Bala e al. 3. D’Amb os, M., Lanza, M.: A Flexible F amewo k o Suppo Collabo a i e So - wa e E olu ion Analysis. In: So wa e Main enance and Reenginee ing. pp. 3–12 (2008) 4. an Dongen, B.F., Van de Aals , W.M.: A Me a Model o P ocess Mining Da a. EMOI-INTEROP 160, 30 (2005) 5. an Dongen, B.F., de Medei os, A.K.A., Ve beek, H., Weij e s, A., Van De Aals , W.M.: The P oM amewo k: A new e a in p ocess mining ool suppo . In: Ap- plica ions and Theo y o Pe i Ne s 2005, pp. 444–454. Sp inge (2005) 6. Feld , R., S a on, M., Hul , E., Liljeg en, T.: Suppo ing so wa e decision mee - ings: Hea maps o isualising es and code measu emen s. In: 39 h Con . on So wa e Enginee ing and Ad anced Applica ions, pp. 62–69. IEEE (2013) 7. Hou, Q., Ma, Y., Chen, J., Xu, Y.: An Empi ical S udy on In e -Commi Times in SVN. In: In . Con . on So wa e Eng. and Knowledge Eng., pp. 132–137. (2014) 8. Kagdi, H., Yusu , S., Male ic, J.I.: Mining Sequences o Changed- iles om Ve sion His o ies. In: Wo kshop on Mining So wa e Reposi o ies. pp. 47–53. ACM (2006) 9. Kindle , E., Rubin, V., Sch¨a e , W.: Ac i i y Mining o Disco e ing So wa e P o- cess Models. So wa e Enginee ing 79, 175–180 (2006) 10. Kindle , E., Rubin, V., Sch¨a e , W.: Inc emen al Wo k low Mining Based on Doc- umen Ve sioning In o ma ion. In: Li, M., Boehm, B., Os e weil, L. (eds.) Uni ying he So wa e P ocess Spec um, LNCS 3840, pp. 287–301. Sp inge (2006) 11. Ogawa, M., Ma, K.L.: So wa e e olu ion s o ylines. In: P oceedings o he 5 h in e na ional symposium on So wa e isualiza ion. pp. 35–42. ACM (2010) 12. OMG: BPMN 2.0. Recommenda ion, OMG (2011) 13. Pila o, C.M., Collins-Sussman, B., Fi zpa ick, B.W.: Ve sion con ol wi h sub e - sion. O’Reilly Media, Inc. (2008) 14. Poncin, W., Se eb enik, A., an den B and, M.: P ocess mining so wa e eposi o- ies. In: So wa e Main enance and Reenginee ing (CSMR), 2011 15 h Eu opean Con e ence on. pp. 5–14. IEEE (2011) 15. Rubin, V., G¨un he , C.W., Van De Aals , W.M., Kindle , E., Van Dongen, B.F., Sch¨a e , W.: P ocess mining amewo k o so wa e p ocesses. In: So wa e P ocess Dynamics and Agili y, pp. 169–181. Sp inge (2007) 16. Rubin, V., Lomazo a, I., an de Aals , W.M.: Agile de elopmen wi h so wa e p ocess mining. In: In . Con . on So w. and Sys em P ocess. pp. 70–74. (2014) 17. To alds, L., Hamano, J.: Gi : Fas e sion con ol sys em. h p://gi -scm.com (2010) 18. Ve beek, H., Buijs, J.C., Van Dongen, B.F., Van De Aals , W.M.: Xes, xesame, and p om 6. In: In o ma ion Sys ems E olu ion, pp. 60–75. Sp inge (2011) 19. Voinea, L., Telea, A.: An Open F amewo k o CVS Reposi o y Que ying, Analysis and Visualiza ion. In: In e na ional Wo kshop on Mining So wa e Reposi o ies (MSR ’06). pp. 33–39. ACM (2006) 20. Voinea, L., Telea, A.: Mul iscale and Mul i a ia e Visualiza ions o So wa e E o- lu ion. In: Symposium on So wa e Visualiza ion. pp. 115–124. ACM (2006) 21. Wilson, J.M.: Gan cha s: A cen ena y app ecia ion. Eu opean Jou nal o Ope - a ional Resea ch 149(2), 430–437 (2003) 22. Wu, J., Spi ze , C., Hassan, A., Hol , R.: E olu ion Spec og aphs: isualizing punc ua ed change in so wa e e olu ion. In: Wo kshop on P inciples o So wa e E olu ion. pp. 57–66 (Sep 2004) 23. Ying, A., Mu phy, G., Ng, R., Chu-Ca oll, M.: P edic ing Sou ce Code Changes by Mining Change His o y. IEEE T ans. So w. Eng. 30(9), 574–586 (2004) 24. Zimme mann, T., Weisge be , P., Diehl, S., Zelle , A.: Mining Ve sion His o ies o Guide So wa e Changes. In: In . Con . So wa e Enginee ing. pp. 563–572. (2004)