Information Technology Reference
In-Depth Information
Process Model Generation from
Natural Language Text
Fabian Friedrich
1
, Jan Mendling
2
, and Frank Puhlmann
1
1
inubit AG, Schoneberger Ufer 89-91, 10785 Berlin, Germany
{
Fabian.Friedrich,Frank.Puhlmann
}
@inubit.com
2
Humboldt-Universitat zu Berlin, Unter den Linden 6, 10099 Berlin, Germany
jan.mendling@wiwi.hu-berlin.de
Abstract.
Business process modeling has become an important tool for
managing organizational change and for capturing requirements of soft-
ware. A central problem in this area is the fact that the acquisition of as-is
models consumes up to 60% of the time spent on process management
projects. This is paradox as there are often extensive documentations
available in companies, but not in a ready-to-use format. In this pa-
per, we tackle this problem based on an automatic approach to generate
BPMN models from natural language text. We combine existing tools
from natural language processing in an innovative way and augmented
them with a suitable anaphora resolution mechanism. The evaluation of
our technique shows that for a set of 47 text-model pairs from industry
and textbooks, we are able to generate on average 77% of the models
correctly.
1
Introduction
Business process management is a discipline which seeks to increase the e-
ciency and effectiveness of companies by holistically analyzing and improving
business processes across departmental boundaries. In order to be able to ana-
lyze a process, a thorough understanding of it is required first. The necessary
level of insight can be obtained by creating a formal model for a given business
process.
The required knowledge for constructing process models has to be made ex-
plicit by actors participating in the process [1]. However, these actors are usually
not qualified to create formal models themselves [2]. For this reason, modeling
experts are employed to iteratively formalize and validate process models in col-
laboration with the domain experts. This traditional procedure of extracting
process models involves interviews, meetings, or workshops [3]. It entails con-
siderable time and costs due to ambiguities or misunderstandings between the
involved participants [4]. Therefore, the initial elicitation of conceptual models
is considered to be a knowledge acquisition bottleneck [5]. According to Herbst
[1] the acquisition of the as-is model in a workflow project requires 60% of the
total time spent. Accordingly, substantial savings are possible by providing ap-
propriate tool support to speed up the acquisition phase.
