Information Technology Reference
In-Depth Information
However, the scope of existing studies is limited, because the metrics used in
these studies assign single (global) values to a BPM to describe its complexity.
Ananda et al. [9] state: “Although studying the overall comprehensibility of
a model is important, from a language evolution perspective it is even more
relevant to discover
which
elements of a notation work well and which do not.”
With this paper, we want to give some first answers on the question
which
relations between model elements in a BPM are dicult to understand.
Despite increasing consciousness about the need to consider comprehensibility
of process models, little research has been undertaken in order to improve and
understand the relationships between modeling elements and comprehensibility.
In this paper, we want to explore comprehensibility as a
local
property. This
means that we measure the comprehensibility of a specific part of a BPM instead
of the model as a whole. This way, we seek to investigate, which relations between
elements in a graphical BPM are dicult to understand. In the research area of
software complexity metrics, similar research has been published by Yang et al.
[10]. Research results suggest that local complexity metrics could be a promising
predictor for understandability. Therefore, we address the question as to when
or under what circumstances similar relationships with local metrics will emerge
in the context of BPM.
Our motivation is to complement the existing stream of work on improving
the comprehensibility of BPM by examining local comprehensibility. In contrast
to existing research such as [7,8] we assign own metrics to each comprehension
question in our study, not to the models as a whole.
The remainder of this paper proceeds as follows: First, comprehensibility of
process model elements and their relationships is placed in context with a review
of relevant theoretical perspectives. Next, we articulate a research model. Then,
we discuss design and findings of an empirical study. The final section discusses
the limitations of our work and presents the implications of our research.
2 Measuring the Cognitive Load of Process Models
2.1 Comprehensibility and Cognitive Load Theory
For defining the term
comprehensibility
, we adapt the definition for understand-
ing of computer programs given by Biggerstaff [11] by relating it to the modeling
context and replacing the word “program” by “BPM”:
“A person understands a BPM when they are able to explain the BPM, its
structure, its behavior, its effects on its operational context, and its relationships
to its application domain in terms that are qualitatively different from the tokens
used to construct the BPM in a modeling language.”
Further popular explanations of the term
comprehensibility
such as “the ease
with which the
...
model can be understood” [12] suggest that cognitive effort
is an important factor determining model comprehensibility and should be as
low as possible. Based on the complex relationships and control flow logic of
organizational processes in practice, understanding of BPM is a task likely to
demand high cognitive effort.
