An Architecture for Querying Business Process, Business Process Instances, and Business Data Models

An A chi ec u e o Que ying Business P ocess, Business P ocess Ins ances, and Business Da a Models Ma ´ıa Te esa G´omez-L´opez1, An onia M. Reina Quin e o1 , Luisa Pa ody1 ,Jose´Miguel P´e ez ´Al a ez1 , and Man ed Reiche 2 1Depa amen o de Lenguajes y Sis emas In o m´a icos, Uni e sidad de Se illa, Se ille, Spain {may egomez, einaqu,lpa ody,josemi}@us.es 2Ins i u e o Da abases and In o ma ion Sys ems, Ulm Uni e si y, Ulm, Ge many [email p o ec ed] Abs ac . Business da a a e usually managed by means o business p ocesses du ing p ocess ins ances. These iewpoin s (business, ins ances and da a) a e s ongly ela ed because he li e-cycle o business da a objec s need o be aligned wi h he business p ocess and p ocess ins ance models. Howe e , cu en app oaches do no p o ide a mechanism o in eg a e hese h ee iewpoin s no o que y hem all oge he while main aining he in o ma ion in he dis ibu ed, he e ogeneous sys ems whe e hey ha e been c ea ed. In his pape , we p opose he in eg a ion o he business p ocess, business p ocess ins ance, and business da a models by using hei me amodels and also an a chi ec u e o suppo his in eg a ion. The goal o his in eg a ion is o make he mos o he h ee models and he echnologies ha suppo hem in an isola ed way. In ou app oach, i is no necessa y o change he sou ce da a o ma s no ans o ming hem in o a common one. Fu he mo e, he p oposed a chi ec u e allows us o que y he h ee models e en hough hey come om h ee diffe en echnologies. Keywo ds: Business Da a Model ·Business P ocess Model Business p ocess ins ance model ·Model in eg a ion He e ogeneous da a sou ces 1 In oduc ion The olume, a ie y and eloci y o p ocess- ela ed da a a e g owing d as ically. Some o hese da a a e c ea ed and managed by Business P ocess Managemen Sys ems (BPMS). A business p ocess (BP o sho ) consis s o a se o ac i i- ies whose execu ion needs o be coo dina ed in an o ganisa ional and echnical en i onmen in o de o achie e a pa icula business goal [24]. The coo dina ed execu ion o ac i i ies is a undamen al p inciple om he iewpoin o Busi- ness P ocess Managemen , and i p o okes he change o da a objec s du ing p ocess ins an ia ion. In gene al, a BP can be implemen ed and ope a ionalized by a BPMS. To make he mos o he he e ogeneous in o ma ion p o ided by he diffe en iewpoin s (i.e., business p ocess, business p ocess ins ance and business da a), i is necessa y o in eg a e he diffe en models and p o ide a mechanism o que y hem all oge he , since hey a e equen ly suppo ed by diffe en echnologies. P e ious solu ions based on seman ic models [10]ha e p o ided mechanisms o c ea e homogeneous da a s o es, using he in o ma ion om he e ogeneous sou ces. The p oblem o his ype o app oaches is ha hey go+ agains he equi emen s es ablished when he e a e a high olume, a ie y and eloci y o da a, as in Big Da a scena ios, whe e he changeabili y and quan- i y o da a make no possible ans o m he sou ces. Fo his eason, his pape p oposes an a chi ec u e whe e p ocess que ying mechanisms a e inspi ed in he Map-Reduce pa adigm. Map-Reduce is a p og amming pa adigm ha allows o massi e scalabili y ac oss hund eds o housands da a nodes, and i le s: (1) di ide he que y in o he subsys ems ha con ain he da a, and (2) me ge he ou pu s o he dis ibu ed da abases. In o de o define how o di ide he que y and combine he ob ained in o ma- ion, i is necessa y o de e mine he ela ion be ween subsys ems. We p opose he use o me amodels and he combina ion o hem. A me amodel defines a ame and a se o ules o c ea ing models o a specific applica ion domain. A iewpoin is defined in ela ion o one o mo e me amodels [9]. In ou scena io, each iewpoin is suppo ed by a me amodel ha defines he main elemen s managed in each case and hei ela ionships. In gene al, he e exis s a un- damen al ela ion be ween he h ee models: Business P ocess Model (BPM), Business P ocess Ins ances Model (BPIM), and Business Da a Model (BDM). Acco dingly, i is no sufficien o que y each o hese models sepa a ely, bu o p o ide in eg a ed access o he p ocess, da a and ins ance pe spec i es. In pa icula , p ocess-cen ic que ies ac oss hese diffe en iewpoin s need o be enabled [4]. The insufficien unde s anding o he inhe en ela ionships exis ing be ween business p ocesses and business da a is deficien [21]. As a as we know, exis ing que ying app oaches ocus on one specific model ype, bu hey do no exploi he in o ma ion esul ing om ha in eg a ion. This pape shows how o in eg a e he BPM, BPIM, and BDM by means o a wea ing model ha makes explici he ela ionships be ween he elemen s o he h ee models. As an example o a BP conside s he o ganisa ion o a con e ence. The co - esponding BP model is shown in Fig. 1. Th ee diffe en pools co e he main unc ions needed o o ganize he con e ence, o submi a pape , and o egis e o he con e ence: (1) he fi s pool desc ibes how o manage he con e ence o ganisa ion by he con e ence chai who is in cha ge o he con e ence equi e- men s; (2) he second pool shows he submission p ocess om he iewpoin o an au ho ; (3) he hi d pool deals wi h he con e ence egis a ion and ela ed paymen . Usually, he execu ion o a p ocess ins ance is pe sis ed in he da abase o he BPMS o which he BP is deployed. Acco ding o he amewo k p oposed in [20], his si ua ion co esponds o a p ocess eposi o y composed o simula ion Fig. 1. Business p ocesses o con e ence o ganisa ions Fig. 2. Concep ual model o he con e ence p ocess models (see De . 3.4 in [20]) This kind o in o ma ion mus ollow he BP ins ance acco ding o he used BPMS. Mo eo e , business da a ha flows h ough he p ocess is pe sis ed in a da abase. Fo he con e ence example, he co esponding concep ual da a model is depic ed in Fig. 2. I p esen s he en i ies in ol ed in he con e ence p ocess. A Con e ence is defined by i s id,name,loca ion,scope (e.g., “So wa e Enginee ing”, “Business P ocess Managemen ”, o “Big Da a”), s a and end da es, and he da e un il which he ea ly egis a ion ee is alid. In u n, an au ho may submi se e al pape s, which a e going o be e iewed by he P og am Commi ee ha decides on he a ing and s a us o he pape . Rega ding he con e ence scena io, he h ee models (i.e., BPM, BPIM and BDM) play an impo an ole in ex ac ing ele an in o ma ion abou he busi- ness p ocess. Examples o que ies making use o he in eg a ion o hese iew- poin s a e “A e age o execu ion ime o he con e ences wi h scope Business P ocess ”, and “E olu ion o he numbe o egis a ions a iable o he p o- cess named ‘Con e ence Managemen P ocess’ o hose con e ences aking place in Ba celona”. In pa icula , his pape shows how he h ee models may be in eg a ed o p ocess such que ies. The pape is o ganized as ollows: Sec . 2in oduces he me amodels ela ed o he business p ocess, business p ocess ins ance and business da a iewpoin s as well as he me amodel ha suppo s he in eg a ion o hem all. A e ha , Sec . 3p esen s ou p oo -o -concep p o o ype. Sec ion 4discusses ela ed wo k. Finally, Sec . 5concludes he pape . 2 Business Viewpoin s This sec ion desc ibes he me amodels co e ing he h ee iewpoin s on a BP, i.e., BPM, BPIM, BDM as well as he me amodel used o in eg a e hem all. Then, hese me amodels a e used o enable an in eg a ion o he iewpoin s. A BP co esponds o a se o coo dina ed ac i i ies, ca ied ou manually o au oma ically, o achie e a specific business goal. The me amodel o c ea ing BP models is summa ised in Sec . 2.1. In u n, he me amodel ela ed o he p o- cess ins ance iewpoin is p esen ed in Sec . 2.2. Finally, conce ning o he da a iewpoin , a simplified me amodel o Unified Modeling Language (UML) [15]is in oduced in Sec . 2.3. In pa icula , a BP needs o con ol i s da a flow, which eflec s he business da a managed in he con ex o he BP and he way his da a is ans e ed h ough he ac i i ies. In gene al, his p ocess- ela ed da a can be desc ibed by a concep ual model which usually is defined using UML. In p ac ice, he h ee models mus be conside ed in an in eg a ed way due o hei many in e dependencies; e.g., BP ins ances a e ela ed o a BP model, da a en i ies a e used o define he da a flow o a BP, and he a ious BP ins ances c ea e da a objec s o upda e he alues o he da a. 2.1 Viewpoin 1: Business P ocess Me amodel The main s anda d used o model business p ocesses is he Business P ocess Model and No a ion (BPMN) as p oposed by OMG [14]. Besides o he elemen s, BPMN 2.0 includes da a objec s, e en s o a ious ypes, and a e ac s. Figu e 3 in oduces a simplified e sion o he BP me amodel (BPMM o sho ), which includes he main me aclasses used in his pape o model he BPs as well as hei ela ionships and a ibu es. The oo o he me amodel is BPMNP ocess, which is composed o a se o BPMNElemen .ABPMNElemen , in u n, may be a Swimlane,a Connec ingObjec ,aFlowObjec ,andanA i ac . No e ha hese me aclasses a e abs ac . The conc e e me aclasses a e sub ypes o hese ones. Thus, a Connec ingObjec may be a MessageFlow,aSequenceFlow, and an Associa ion. Fu he mo e, he e a e h ee ypes o FlowObjec : E en ,Ga eway and Ac i i y. The h ee o hem a e also abs ac me a- classes. An E en may be an Ini ialE en ,anIn e media eE en , and a FinalE en .AGa eway may be a XOR,OR,andAND ga eway. An Ac i i y Fig. 3. Simplified business p ocess me amodel may be a Task (also an abs ac me aclass) o a Subp ocess. The e a e diffe - en ypes o Task:ManualTask,Recei edTask,SendTask,Sc ip Task, Se iceTask,andUse Task. No e ha he seman ics o hese me aclasses is exac ly he specified by he BPMN2.0 s anda d and also ha his me amodel is a simplified e sion o i . We e e in e es ed eade s o [14] o knowing he seman ics o each me aclass. 2.2 Viewpoin 2: Business P ocess Ins ance Me amodel A business p ocess ins ance ep esen s a conc e e case in he ope a ional business o a company [24]. Figu e 4shows ou p oposal o Business P o- cess Ins ance Me amodel (BPIMM o sho ). P ocessEngine is he oo o he me amodel and i ep esen s he Business P ocess Managemen Sys- em (BPMS) ha execu es he BP. A BPMS may deploy a ious p ocesses. Each p ocess is desc ibed by means o he P ocessDe ini ion me a- class. No e ha P ocessDe ini ion is ela ed o a Business P ocess (c . Sec . 2.1). P ocessIns ance ep esen s an execu ion o a specific p ocess. Thus, a P ocessDe ini ion is ela ed o a se o P ocessIns ance. AP ocessDe ini ion is composed o a sequence o Ac i i y. Acco d- ingly, a P ocessIns ance is composed o a se o Ac i i yIns ance (childAc i i ies ela ion). The e is also a pa en -child ela ion be ween ac i i ies ins ances. The fi s ac i i y ha is ins an ia ed du ing a p ocess exe- cu ion is an ac i i y ins ance ha has no pa en . Finally, a se o Va iable can be associa ed o a P ocess Ins ance. No e ha BPIMM deals wi h concep s ela ed o p ocess execu ion and ha he defini ion o p ocesses included in many BPMSs usually has mo e p ope - ies o p ocesses and ac i i ies han he s anda d BPMN 2.0. Ha ing his in o accoun , he P ocessDefini ion and Ac i i y me aclasses model hese p ope ies ha a e p esen in he execu ion en i onmen s. 2.3 Viewpoin 3: Business Da a Me amodel In o de o be able o c ea e BDM, we use he OMG Me amodel o ep e- sen Concep ual Models in oduced in Fig. 5, which includes he main en i- ies o he Business Da a me amodel (BDMM) p esumed in his pape . Fig. 4. Business P ocess Ins ance Me amodel (BPIMM) Fig. 5. Simplified Business Da a Me amodel (BDMM) The oo en i y is BMDiag am, which is composed o a se o ModelElemen . AModelElemen may be an Associa ionEnd,aRela ionship,a Classi ie ,aTypedElemen ,o aFea u e. The en i ies o he sys em a e mainly ep esen ed by a Classi ie , ei he an Associa ion,Class,o Da aType. Fu he mo e, h ough he Associa ionEnd and Rela ionship, Classi ie en i ies a e ela ed. No e ha an ins ance o his me amodel which depic s en i ies ela ed o he o ganisa ion o a con e ence is p esen ed in Fig. 2. 2.4 In eg a ing BPM, BPIM and BDM Business p ocess, business p ocess ins ance and business da a me amodels ep- esen diffe en , bu complemen a y, iewpoin s o a business p ocess. Linking en i ies ac oss hese me amodels is a mus in o de o exploi he in o ma ion p o ided by hei specific ins ances, i.e., hei models. In his con ex , model wea ing is a gene ic ope a ion ha es ablishes co espondences be ween model elemen s. The esul ing wea ing models a e a special kind o models ha link oge he o he models. In gene al, wea ing models con ain a se o links be ween he elemen s o wo diffe en models [8]. A wea ing model con o ms o a wea ing me amodel, which defines he kind o links ha may be es ablished be ween he elemen s o he wo en models. A las Model Wea e (AMW) [8] is a ool ha allows c ea ing and handle wea ing models and me amodels. New wea ing me amodels should be imple- men ed by ex ending a Co e Wea ing Me amodel ha i is included and imple- men ed in AMW. Thus, o define ou in eg a ion me amodel we ex end his Co e Wea ing Me amodel. Figu e 6shows ou ex ension (see he me aclasses wi h g ey backg ound colou ), which specifies he seman ics o he links be ween he elemen s o Business P ocess, Business P ocess Ins ance and Business Da a models. These elemen s a e specific ins ances o he co esponding me amodels (i.e., BPMM, BDMM, and BPIMM). No e ha classes belonging o he co e wea ing me amodel a e abs ac , and a e ex ended by me aclasses ( he ones colou ed in g ey) ha efine hem by adding he seman ics needed in he con- ex o ou scena io. An In eg a ion Model can be seen as a kind o wea ing model ha le us in eg a e he h ee kinds o models (BPIM, BDM, and BPM). This in eg a ion model defines he ela ionships be ween he elemen s in he h ee models. Ma ch is a kind o link ep esen ing he ela ion be ween wo elemen s om diffe en models. Figu e 7shows how h ee ins ances (i.e., models) o BPIMM, BPMM and BDMM may be ela ed by means o an ins ance o he In eg a ion Me a- model. Classes wi h <<BPIM>> s e eo ype ep esen ins ances o Business P o- cess Ins ance me aclasses, classes wi h <<BPM>> ep esen ins ances o Business Fig. 6. In eg a ion me amodel Fig. 7. In eg a ed BPIM, BPM and BDM P ocess me aclasses, classes wi h <<BDM>> s e eo ype ep esen ins ances o Business Da a me aclasses, and classes wi h <<IM>> ep esen ins ances o In eg a ion me aclasses. Thus, he Con e enceManagemen ::BPMNP ocess ins ance ep esen s he p ocess named Con e ence Managemen P ocess, as depic ed in Fig. 1, whe eas he ins ance ia1::Ac i i yIns ance ep esen s one execu ion o he ac i i y ha configu es he Con e ence BPM17. 3 A chi ec u e o Combined Que ies Da a combina ion is an impo an open p oblem in business in elligence, as explained in [20], whe e he au ho s p opose a gene al, abs ac amewo k o de ising p ocess que ying me hods. Ou a chi ec u e, based on Map-Reduce pa adigm, defines how he que y is di ided (Map), and he pa ial que ies a e combined (Reduce). Ou p oposal can be seen as an ins ance o he abs ac amewo k p esen ed in [20], in such a way ha i includes a eal combina ion o echnologies ha can be hough as he implemen a ion o some o he gene ic unc ionali ies desc ibed in he amewo k. Assume one wishes o know how many ins i u ions a e in ol ed in he execu ion o each ask, in a con ex whe e he BPMS Boni aTM, da a a e pe sis ed in an O acle da abase and he BPM is pe sis ed by using Neo4J a que y simila o he one shown in Lis ing 1should be execu ed: SELECT Task.id, COUNT(ins i u ion) FROM (Task FULL JOIN Ac i i yIns ance FULL JOIN Pe son) GROUP BY Task.id WHERE Task.idP ocess = idP Lis ing 1. Que y example No e ha ob ain he que y esul s, you need o que y da a ela ed o he h ee iewpoin s. In o he wo ds, you need in o ma ion abou Tasks om he BP iewpoin , in o ma ion abou Ac i i y Ins ances om he BPI iewpoin , and in o ma ion abou Pe sons om he BD iewpoin . No e ha o ob ain his in o ma ion we ha e o deal wi h h ee diffe en echnologies (Neo4J, Boni a, and O acle, espec i ely). We p opose a wo-s ep p ocess o execu e he que y aligned wi h Map-Reduce me hods. The fi s s ep, da a ex ac ion, is he esponsible o ans o ming he o iginal que y in o a se o que ies ha a e specific o each iewpoin . The second s ep, da a in eg a ion, has as inpu he esul s o he p e ious que ies and is he esponsible o combining hem. Figu e 8shows he a chi ec u e ha suppo s he wo-s eps que y execu ion. The ollowing sec ions gi e some de ails abou hese wo s eps. 3.1 Da a Ex ac ion In he da a ex ac ion s ep, he o iginal que y is ans o med in o h ee diffe en que ies, in o de o que y he h ee diffe en iewpoin s in an isola ed way (see (1) in Fig. 8). The esul o hese specific que ies (s ep 2) is a se o JSON files ha will be combined in s ep (3). The in e ac ion wi h he specific echnology ha suppo s each iewpoin (i.e., Neo4J, Boni a and O acle) is in cha ge o d i e s. D i e s a e he componen s ha deal wi h he pla o m specific ea u es o each echnology. In o he wo ds, d i e s ans o m a gene ic que y o a specific que y on he espec i e da a eposi o y. No e ha hese h ee que ies can be execu ed in pa allel since each esul is ob ained om isola ed da abases. All d i e s should implemen he gene ic ope a ion selec . This selec ope - a ion ex ac s a se wi h all o he objec s o a gi en class ha ulfil ce ain p edica e. Tha is, e e y d i e should implemen an ope a ion wi h a p o- o ype simila o he ollowing one: Se <T>selec (Class<?>name, P edica e<T>p ed). Thus, ou o iginal que y (see Lis ing 1) is ans o med in o h ee selec calls (one o ob ain a se o asks, ano he one o ob ain a se o Ac i i y Ins ances, and, finally, one o ob ain a se o Pe sons) ha will be execu ed by one d i e each. The Neo4J D i e ans o ms he selec ope a ion in o a Neo4J que y o ob ain he se o asks o he p ocess whose id is idP. The d i e uses Sp ing Da a and neo4j-ogm-bol -d i e o map he g aph da abase in o Ja a Objec s. The Boni a D i e ans o ms he selec ope a ion in o a se o calls o he Fig. 8. Combina ion o echnologies o he a chi ec u e