Welcome to your wiki!

This is the default page, edit it as you see fit. To add a new page simply reference it within brackets, e.g.: [SamplePage].

The wiki uses Markdown syntax.

Process Extractor

The process extractor mines business process models from event logs. Various extensions have been developed to enable use-case specific mining, as well as filtering and noise reduction.

Usage Preconditions

The tool is a standalone Java application which takes a text file as input. In order for the process extractor to work, the content of the text file needs to comply to one of the supported log formats. We support a log format developed in the project, which is an extension of the XES event log standard, as well as logs created with the help of the open source library OpenXES. Furthermore, an adapter has been created within the process extractor that allows for easy transformation of domain specific logs into the required format. This, for instance, enables the application of the process extractor in the context of the LNEC sensor use case by reading and transforming files extracted from an event log database table.

When using the process extractor, the user may provide several parameters to the tool:

• input file path: the path to the file containing the event log that is to be mined
• output file path: the path where the process extraction result is written to
• optimisation duration: determines how long the process extraction should run
• output type: the type of the output model; this may be either "--bpmn" or "--archi"
  • --bpmn: a BPMN model is generated that uses the iERM format
  • --archi: an Archimate model is generated, which may be viewed and edited with Archi
• debug: with this optional parameter debug mode can be enabled which informs the user about the process extraction progress, including screenshots of the extracted process models

If the extractor is used incorrectly, the process extractor provides the user with instructions on how to use it correctly.

Output Description

The output of the process extractor is a process model in BPMN format, along with performance information that has been mined from the event log. This format may be imported directly into iERM, or viewed with the visualisation capabilities built into the process extractor.
As an alternative option, the result can be transformed into the Archimate model format, from which it may be further transformed into the TIMBUS context model by applying the archi2owl transformer tool.

Use Cases

Use Case 1

When running the process extractor on the sensor use case of WP8, a file in CSV format created from tables in the sensor database serves as input. This file contains time stamps for sensor data acquisition as well as information about the access of this data by users for transformation and analysis of the data.
If called from command line, the command to extract process model from WP8 looks as follows:

java -jar ./tools/process-miner.jar "./logs/sensor.csv" "./models/wp8.bpmn" 20 --bpmn --debug

This call initiates a transformation of the WP8 internal log format into XES format. Thereafter, the process extractor mines a BPMN model from the event log. The result can be seen below:

In this case the transformation and preprocessing adapters have been set up so they interpret only the abstract activities relating to sensor data showing in the event log. These are the activities of adding and deleting campaigns for sensor data acquisition, as well as transforming, analysing and saving these data sets. Note that the model shown in the picture is the result of a very early extraction without preprocessing of the data or an optimised configuration of the extractor for the use case. This, as well as an evaluation of the tool in connection with the use case will be performed in near future when an extensive amount of log data has been gathered.

Use Case 2

In the case of WP9, where logs are in the the standardised XES format, the process extractor is called like this:

java -jar ./tools/process-miner.jar "./logs/drugfusion.xes" "./models/drugfusion.archimate" 20 --archi

The result can be seen below:

This result is an Archimate model, which may be used to transform into the TIMBUS context model. For a direct import into iERM, the BPMN model is generated (as via the call listed in the WP8 example) which also includes performance information required by the simulation engine in iERM, thus not requiring further manual annotation.
As in the example of WP8 discussed above, the shown result proves technical feasibility and applicability of the process extractor to the use case, but will need to be optimised further.

Extension: Process Monitor

As opposed to the Process Extractor the Process Monitor is working in a continuous fashion and not as a static tool. It monitors an event stream and processes the events into an "online" process footprint in a dynamic fashion. In a second step this footprint can then be interpreted on-demand or at run-time into a business process model conforming to BPMN terminology. Additionally to the evolutionary approach that could be used for the footprint interpretation, a deterministic approach called Constructs Competition Miner has been developed and complies to additional execution time constraints and is thus more suitable for process monitoring at run-time. If a change in the process execution is detected a notification could be generated which in turn may trigger a the preservation process.

Usage Preconditions

Using Webservice RESTful API an interface is provided which can be registered to. This interface provides a method called onEvent(Event e) into which each event needs to be passed (each with its own method call) and processed by the process monitor. The event has to conform to the same event format the process extractor works with (see above - extension of the XES format).

Output Description

The continuously monitored process is available in-memory in the process monitor Java application. With the function BPModel getCurrentModel(), again provided by the interface (Webservice RESTful API) the model can be extracted on-demand and if necessary be transformed into an archimate model. Alternatively, the model can be directly imported by the iERM tool in order to allow for additional analysis.

Use Cases

Since the use case from WP9 supports a change of the business process model and is thus best suited, the process monitor is show-cased and integrated into the WP9 use case.

Changelog

• conformance checking methods have been added to evaluate extractor result
  • replay algorithm on BPMN models to check trace compliance
  • footprint extraction from both model and log, and their comparison in order to accelerate the evolutionary computation based approach used in the process extractor
• adaptation of the interface model to iERM
• created adapter for reading and transforming WP8 event logs
• created adapter for visualising WP8 and WP9 event logs
• created adapter for transforming the output into Archimate format.
• created process monitor