Monitoring & Controlling · November 17, 2021

Colliery

COLLIERY (COLLaboration dIscovERY) is a novel technique that permits discovering a business process collaboration from a set of event logs of a distributed system. The technique is parametric with respect to the algorithm to be used for process discovery. In this way, we can take advantage of already available algorithms introduced by the process mining community. Specifically, given the logs belonging to the organizations that participate in a collaboration, the selected process mining algorithm separately discovers each organization’s processes. Then, by analyzing the logs for retrieving information about message exchanges, the processes are automatically combined to form a business process collaboration diagram representing the distributed system’s overall behavior.

USER GUIDE

Prerequisites:

Make sure to have downloaded in the same directory both the Colliery jar file and the scripts folder. Then execute the jar file with a double click. Some examples are provided here.

The COLLIERY technique is structured in the following phases. In the Logging phase, while the distributed system under consideration is in execution, each system participant independently and locally logs the events concerning its execution, including the exchange of messages with the other participants. In the Processes Discovery phase, the log produced by each participant is separately processed using the process mining algorithm selected by the COLLIERY’s user. Indeed, the technique is parametric to the algorithm used in this phase, which is the same for all logs of the system. This parameterization enables the usage of already available algorithms introduced by the process mining community. This phase’s result is a BPMN process diagram for each log taken as input. In the Messages Analysis phase, the information concerning the exchanges of messages stored in the logs is retrieved and analyzed. The aim of this phase is twofold. Firstly, it produces information on communication aspects necessary to properly build a collaboration diagram from the discovered processes. Secondly, analytics on message delivery and consumption, and related time, are produced to help the user identify potential issues in the collaboration. In the Collaboration building phase, the BPMN collaboration diagram representing the overall system is built as follows. Each previously generated BPMN process is inserted in a pool. The information on communication produced by the previous phase is exploited to properly connect activities and events of different processes with message flows. Such information is also used to tailor the processes to keep into account those features, distinctive of collaborations, that are neglected by the process mining algorithm (e.g., event-based gateways and message events).