Tools and Resources

SemTraE2B is a tool for semi-automatically translating EPC process models into BPMN. It is based upon a set of translation rules that map elements of a EPC meta-model to elements of the BPMN meta-model. An additional configuration offers the possibility to choose more specialized model elements (like for example specific event or task types) for the resulting BPMN diagram.

This section provides both, a screencast that shows the tool as well as a link to the web-based software itself where you can try it by yourself.

Link to the tool: SemTraE2B

Flexible business processes can often be represented more easily using a declarative process modeling language (DPML) rather than an imperative language. Process mining techniques can be used to automate the discovery of process models. One way to evaluate process mining techniques is to synthesize event logs from a source model via simulation techniques and to compare the discovered model with the source model. Though there are several declarative process mining techniques, there is a lack of simulation approaches. Process models also involve multiple aspects, like the ow of activities and resource assignment constraints. MuDePS is an approach that automatically synthesizes event logs that conform to a given model specified in the multi-perspective, declarative language DPIL. Our technique translates DPIL constraints to a logic language called Alloy. A formula-analysis step is the actual log generation.

The tool is implemented as a RapidMiner plug-in. In order to get a brief impression of how to use MuDePS within RapidMiner, please watch the video below.

In order to use the tool, you can either use the pre-compiled jar or (modify and) compile the sources by yourself. The jar must be placed in the extension directory of RapidMiner. Hence, it is also necessary that you have RapidMiner () installed.

• pre-compiled jar: MuDePS
• sources: MuDePS_sources

We are currently working on an extension that provides support for the well known multi-perspective, declarative process modeling language MP-Declare. Hence, we are still interested in feedback and collaborations related to this topic. Thus, don't hestitate to contact us if you have any questions: lars.ackermann@uni-bayreuth.de.

Business process models are used to visualize processes in order to make processes more comprehensible. In many cases, process models contain the process steps of all involved participants, leading to large and potentially confusing diagrams comprising information that seem unimportant for single participants. To tackle this problem, individual and flexible views on business processes are needed. In this work we present CUPAVIS, a tool to construct parameterizable process views based on BPMN models. Individual and customizable process views from an underlying process model may be obtained by applying adequate abstraction mechanisms which reduce the complexity of the model. Depending on the chosen parameterization, the resulting view is more or less abstract what enables users to determine the loss of information. Furthermore CUPAVIS provides different representation forms of process views to support other perspectives and to reveal hidden process details.

• Link to the tool:  CUPAVIS
• Example BPMN models:  example_models.zip
  

Supplementary materials of the paper "An Experimental Study of Intuitive Representations of Process Task Annotations":

Research Method

• Visualization of the study procedure: study_procedure.pdf

• All used questionnaires:
  
  1. German version (original): questionnaires_de.zip
  2. English version: questionnaires_en.zip

• Further descriptions of the questionnaires and recruitment: questionnaires_descriptions.pdf

• All used process models:
1. German version (original): process_models_de.zip
2. English version: process_models_en.zip

• Study setup: study_setup.pdf

Analysis and Results

• Descriptive statistics: descriptive_statistics.pdf
• Complementing Table for hypothesis 1: table_hypothesis1.pdf
• Complementing Figures for hypothesis 2: figures_hypothesis2.pdf

Natural Language Processing (NLP) is a discipline that bridges computer science and linguistics. Within this discipline typical examples for (so called) downstream tasks are textclassification (e.g. spam detection in emails), sentiment analysis (e.g. extracting an overallrating from product reviews) and information retrieval (e.g. creating chatbots orautomatically summarizing books). Some tasks are divided into several steps solved byseparate techniques, such as data augmentation (for generating artificial data in case ofdata scarcity) or basic feature extraction (e.g. determining word types like nouns, verbs,etc.). Boosted by the rapid progress in Artificial Intelligence (AI) NLP is a rapidly evolvingresearch field. However, most of the scientific breakthroughs can be traced back toresearch groups sponsored or owned by large companies like Google, Amazon or OpenAI.To develop new solutions for the above tasks it is necessary to compare them to relatedapproaches. However, for smaller groups we suspect that it is hard to do so for a couple of issues:

1. Researchers encounter a huge overhead to set up the experiment infrastructure,since they have to deal with, for instance, non-standardized dataset formats,pre-processing, and visualization - tasks that are frequently occurring and areusually solved in the same way as many times before. Thus, experiment setupsshould be identical to those used for evaluating related approaches, but ...
2. usually those setups are heterogeneous in terms of datasets used, metricdefinitions or parameterizations of the investigated techniques. One consequencehere is that it is usually impossible to compare techniques based on a literaturereview only. Hence, each research group has to run their own experiments.
3. Experiments reported in scientific literature are often not reproducible due tomissing setup details, unavailable implementations or
4. hardware requirements, a small research group is often not able to fulfill basedon their own resources.5. Often it is not possible to run experiments because data has to be submitted tonon-EU countries (e.g. when using Google Cloud), which can be an issue in terms of regulations like, for instance, the General Data Protection Regulation (GDPR).

All of the above issues emerged in our own research and, thus, our key users areresearchers of Universities. To publish an article, developed approaches have to be madeavailable in order to allow reviewers (members of a program committee) to easilycomprehend and reproduce experiment results reported in this article.

The above conjectures are currently being investigated by a broad survey via LimeSurvey. If you are a participant and have arrived at this page by clicking on our link provided in the survey, please consider this as confirmation of the survey's authenticity. If you have any questions please contact lars.ackermann [@] uni-bayreuth.de.