Abstract process model refers to a coarse-grained view of a process model. Recognizing the diverse usability of abstract models, a plethora of process model abstraction techniques have been proposed. However, these techniques treat process fragments as black box and replace the fragments for abstraction, without taking into consideration the semantics of activities. Consequently, the Business Significant Activities (BSA) inside the black box are also eliminated, which impedes the robustness of abstraction. To address that problem, in this paper, we have proposed an activity-based approach in which all BSAs, including the ones inside process fragments, are preserved. Specifically, we have employed a systematic and rigorous procedure to develop a benchmark collection of 960 process models. The collection includes 240 source process models and 720 process models abstracted at three levels of granularity. Subsequently, we have evaluated the effectiveness of eight activity-ranking techniques for identifying BSAs. The results show, there is no universal technique that achieve higher accuracy for all types of process models. However, in majority of the cases, Word Frequency, Word Co-occurrence and Label Centrality are the top performing techniques for identifying business significant activities.
[1]
Khurram Shahzad,et al.
Generating process model collection with diverse label and structural features
,
2016,
2016 Sixth International Conference on Innovative Computing Technology (INTECH).
[2]
George D. C. Cavalcanti,et al.
Assessing sentence scoring techniques for extractive text summarization
,
2013,
Expert Syst. Appl..
[3]
Henrik Leopold,et al.
A Textual Description Based Approach to Process Matching
,
2016,
PoEM.
[4]
Mathias Weske,et al.
Business Process Model Abstraction
,
2010,
BPM 2010.
[5]
Mathias Weske,et al.
Business process model abstraction: a definition, catalog, and survey
,
2012,
Distributed and Parallel Databases.
[6]
Heiner Stuckenschmidt,et al.
A probabilistic evaluation procedure for process model matching techniques
,
2018,
Data Knowl. Eng..
[7]
Mathias Weske,et al.
On Application of Structural Decomposition for Process Model Abstraction
,
2009,
BPSC.
[8]
Khurram Shahzad,et al.
Comparing manual- and auto-generated textual descriptions of business process models
,
2016,
2016 Sixth International Conference on Innovative Computing Technology (INTECH).