Traditionally, to monitor the execution of a business process, organizations rely on monitoring modules provided by Business Process Management Systems (BPMSs), which automate and keep track of the execution of processes. While the adoption of a BPMS to monitor a single-party, fully-automated business process is straightforward, the same cannot be said for multi-party processes heavily relying on manual activities. In fact, a BPMS requires explicit notifications to determine when activities that are not under its direct control are executed. This requires organizations to federate their BPMS, a complex task that has to be performed whenever a new organization participates in the process. Also, when activities are not automated, human operators are responsible for manually sending notifications to the BPMS, a task that disrupts the operators' work and, as such, is prone to be forgotten or postponed. To continuously and autonomously monitor multi-party processes involving non-automated activities, this thesis proposes a novel technique, named artifact-driven process monitoring. This technique exploits the Internet of Things (IoT) paradigm to make the physical objects participating in a process smart. Being equipped with sensors, a computing device, and a communication interface, such smart objects can then become self-aware of their own conditions and of the process they participate in, and exchange this information with the other smart objects and the involved organizations. This way, it is possible for the monitoring infrastructure to stay in close contact with the process, and to cross the boundaries of the organizations. To be aware of the process to monitor, instead of using activity-centric process models, usually adopted by BPMSs, smart objects rely on an extension of the Guard-Stage-Milestone (GSM) artifact-centric modeling language, named Extended-GSM (E-GSM). Normally, a BPMS expects the execution to rigidly adhere to the process model defined in advance. Therefore, whenever a deviation between the execution and the model is detected, a BPMS requires human intervention to resume process monitoring. E-GSM, on the other hand, treats the execution flow (i.e., dependencies among activities) in a descriptive rather than prescriptive way. Consequently, smart objects can detect violations during execution without interrupting the monitoring. Additionally, E-GSM can monitor if the physical objects evolve as expected while the process is executed. Finally, E-GSM provides constructs to determine, based on the conditions of the physical objects, when activities are started or ended. This thesis also presents an approach to determine to which extent smart objects are suited to monitor a particular process, given their sensing capabilities. To relieve process designers from learning the E-GSM notation, and to allow organizations to reuse preexisting process models, a method to instruct smart objects given Business Process Model and Notation (BPMN) collaboration diagrams is also presented. Finally, a prototype of an artifact-driven monitoring platform, named SMARTifact, is developed and tested against both historical and live sensor data.
Tradizionalmente, al fine di monitorare l'esecuzione dei propri processi aziendali, le organizzazioni si affidano ai moduli di monitoraggio dei Business Process Management System (BPMS), strumenti software spesso già impiegati per l'automazione di tali processi. Tuttavia, i BPMS mal si prestano al monitoraggio di processi distribuiti tra più organizzazioni, e con una forte presenza di attività non automatizzate. Difatti, un BPMS richiede l'invio di notifiche esplicite per determinare quando un'attività da esso non controllata viene eseguita. Ciò richiede pertanto, nel caso di processi distribuiti, che le organizzazioni partecipanti federino i propri BPMS, compito decisamente complesso e che deve essere ripetuto ogni volta che il processo viene esteso ad una nuova organizzazione. Quando invece sono presenti attività non automatizzate, è compito del personale dedicato allo svolgimento di tali attività inviare al BPMS le notifiche. Tale compito costringe dunque il personale ad interrompere il loro abituale lavoro, e viene pertanto facilmente dimenticato o posticipato. Al fine di poter superare queste limitazioni, questa tesi propone una nuova tecnica, chiamata artifact-driven process monitoring, la quale permette di monitorare in modo continuato ed autonomo processi distribuiti e con presenza di attività non automatizzate. Questa tecnica sfrutta il paradigma dell'Internet of Things (IoT) per rendere intelligenti gli oggetti tangibili che partecipano al processo. Equipaggiando tali oggetti con sensori, un dispositivo di calcolo ed un'interfaccia di comunicazione, è possibile trasformarli in smart object e, così facendo, farli diventare consapevoli delle proprie condizioni e di come il processo al quale partecipano è organizzato, nonché permettergli di comunicare queste informazioni agli altri oggetti intelligenti ed alle organizzazioni facenti parte del processo. In questo modo, l'infrastruttura di monitoraggio può stare a stretto contatto con il processo, ed attraversare i confini delle singole organizzazioni. Per poter conoscere come il processo da monitorare è strutturato, anziché usare modelli di processo activity-centric, solitamente adottati dai BPMS, gli smart object utilizzano un'estensione del linguaggio artifact-centric Guard-Stage-Milestone (GSM), chiamata Extended-GSM (E-GSM). Di norma, un BPMS si aspetta che un processo venga eseguito esattamente secondo quanto riportato in un modello formalizzato anticipatamente. Pertanto, ogni volta che riscontra una discrepanza tra modello ed esecuzione effettiva, esso richiede l'intervento di un operatore per poter continuare il monitoraggio. Al contrario, E-GSM considera il flusso di esecuzione, ovvero le dipendenze tra attività, in modo descrittivo anziché prescrittivo. Di conseguenza, gli smart object sono in grado di rilevare violazioni mentre il processo viene eseguito senza interromperne il monitoraggio. Oltre a ciò, E-GSM permette anche di monitorare se gli smart object vengono manipolati correttamente durante l'esecuzione del processo. Infine, E-GSM fornisce costrutti per definire, in base alle condizioni degli smart object, quando le attività iniziano o terminano. Questa tesi presenta inoltre un approccio volto a quantificare, in base alle capacità della sensoristica, fino a che punto gli smart object risultano essere adeguati al monitoraggio di uno specifico processo. Al fine di sollevare i progettisti di processo dall'apprendimento della notazione E-GSM, e di permettere il riuso di modelli di processo preesistenti, viene inoltre presentato un metodo per configurare gli smart objects partendo dai diagrammi collaborativi Business Process and Notation (BPMN). Infine, è stato sviluppato un prototipo di piattaforma di monitoraggio artifact-driven chiamato SMARTifact, il quale è stato testato utilizzando dati sensoristici sia storici sia in tempo reale.
Artifact-driven business process monitoring
MERONI, GIOVANNI
Abstract
Traditionally, to monitor the execution of a business process, organizations rely on monitoring modules provided by Business Process Management Systems (BPMSs), which automate and keep track of the execution of processes. While the adoption of a BPMS to monitor a single-party, fully-automated business process is straightforward, the same cannot be said for multi-party processes heavily relying on manual activities. In fact, a BPMS requires explicit notifications to determine when activities that are not under its direct control are executed. This requires organizations to federate their BPMS, a complex task that has to be performed whenever a new organization participates in the process. Also, when activities are not automated, human operators are responsible for manually sending notifications to the BPMS, a task that disrupts the operators' work and, as such, is prone to be forgotten or postponed. To continuously and autonomously monitor multi-party processes involving non-automated activities, this thesis proposes a novel technique, named artifact-driven process monitoring. This technique exploits the Internet of Things (IoT) paradigm to make the physical objects participating in a process smart. Being equipped with sensors, a computing device, and a communication interface, such smart objects can then become self-aware of their own conditions and of the process they participate in, and exchange this information with the other smart objects and the involved organizations. This way, it is possible for the monitoring infrastructure to stay in close contact with the process, and to cross the boundaries of the organizations. To be aware of the process to monitor, instead of using activity-centric process models, usually adopted by BPMSs, smart objects rely on an extension of the Guard-Stage-Milestone (GSM) artifact-centric modeling language, named Extended-GSM (E-GSM). Normally, a BPMS expects the execution to rigidly adhere to the process model defined in advance. Therefore, whenever a deviation between the execution and the model is detected, a BPMS requires human intervention to resume process monitoring. E-GSM, on the other hand, treats the execution flow (i.e., dependencies among activities) in a descriptive rather than prescriptive way. Consequently, smart objects can detect violations during execution without interrupting the monitoring. Additionally, E-GSM can monitor if the physical objects evolve as expected while the process is executed. Finally, E-GSM provides constructs to determine, based on the conditions of the physical objects, when activities are started or ended. This thesis also presents an approach to determine to which extent smart objects are suited to monitor a particular process, given their sensing capabilities. To relieve process designers from learning the E-GSM notation, and to allow organizations to reuse preexisting process models, a method to instruct smart objects given Business Process Model and Notation (BPMN) collaboration diagrams is also presented. Finally, a prototype of an artifact-driven monitoring platform, named SMARTifact, is developed and tested against both historical and live sensor data.File | Dimensione | Formato | |
---|---|---|---|
thesis_final.pdf
accessibile in internet per tutti
Descrizione: Testo della tesi
Dimensione
11.14 MB
Formato
Adobe PDF
|
11.14 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/141243