Process modeling techniques play an important role to capture information about business procedures. This paper suggests two novel methods for business process modeling. The first method allows generating process models from process descriptions created with controlled natural language. This method is based on a parser for natural language and is supported by sentence templates and an autocomplete function. The second method suggests a collaborative setting, which allows discovering process models through user interactions. Both methods have been implemented in a prototype. The aim of this paper is to show new possibilities for process modeling through the combination of the two methods.
Today organizations have to adapt and improve their business processes more often and on short notice. Thus, there is a need for effective and efficient methods for business process management. The documentation and modelling of existing processes is an important part in this context; but in many cases the results of process modeling projects (with current methods) do not fully comply with the expectations of the parties involved. To demonstrate new process modeling possibilities, two methods have been developed and implemented within a protoype. The presented tool is an ASP.NET MVC web application executed in a web browser. In addition, there is a Ms Office-App available to have an integration into Office products, which are often used to present or describe process models. The prototype will be available at http://bpm.caporale.eu soon.
The first method presented in this paper, is an approach of generating process
models from natural language text. Instead of analyzing descriptions of business
tasks and then generating the process model (related approaches are presented
in e.g. [
The method is based on so-called sentence templates. Sentence templates
are often used to describe requirements for software development projects. They
can be considered as a support technique, which helps the user formulating
understandable sentences. With the help of these templates, the modelers are
able to describe their business tasks in controlled natural language, which will
be automatically transformed into a process model. For each basic workflow
control-flow pattern (Sequence, Alternative, Parallel Split, Synchronization and
Simple Merge), a sentence template has been defined for the German and English
language. An example can be found in e.g. [
Descriptions that have been constructed using the sentence templates are
automatically analyzed by the tool. For this purpose, the approach of [
When a process modeler uses the tool, he will see a text-box on the left and the sentence templates on the right side of the tool. Synchronously to typing in the natural language text, a process model is generated at the bottom side of the tool and the sentence template is dynamically adjusted to the current context. In addition a recommender suggests possible formulations with respect to the current parser’s state similar to an auto-completion function under the text-box (Fig. 1). 2.2
The second method addresses the modeling process itself. Most of the existing modeling approaches aim to extract the process knowledge of an organization through e.g. expert interviews or workshops, which can be considered as a top down approach. In contrast, the method presented is a bottom up techniques for process discovery. The advantage of this method is to bring process modeling activities closer to the knowledge carriers. Assuming, that the tool for generating process models from natural language text can be used by knowledge carriers successfully, the following adaptions will lead to a novel collaborative setting.
The first adaption is, that the user only describes his own activities and
has to provide information about the precondition and postcondition. A related
approach is [
Whenever the user describes a process using the tool containing such preconditions or postconditions, the system will execute a workflow, which will inform the mentioned persons by e-mail. The e-mails will include information about the just created process and ask the receiver to provide more information to the system by clicking on a specially generated link within the e-mail. Clicking on the link will trigger a validation workflow on server-side and create a new process description for the new user that has been addressed within the e-mail.
When knowledge carriers use this collaborative approach and describe their own activities, the system will store many process models with connections between them. As this structures are similar to event logs, process mining techniques are applicable to discover more general process models. 3
Conclusion and Outlook The presented tool combines a method for generating process models from natural language with a workflow oriented collaborative setting and shows new possibilities for process modeling.
As an outlook we assume, that the workflow oriented approach for business process modeling has several advantages. First, the fact that every knowledge carrier will only describe his own activities and will not make any assumptions about activities from other people could possibly reduce misunderstandings, which occur with other techniques when a process modeler has to understand the peoples’ tasks. Second, the bottom up approach could be a new technique for discovering undescribed and unmentioned processes throughout the organization. It even even represents a new approach on Adaptive Case Management. Last, it is reasonable, that the e-mails send by the workflow tool could possibly cause a chain reaction for a new way of collaborative business process modeling.
The next steps include a first evaluation to get feedback about the described methods and improve the underlying modules and user interface.