As business process management is increasingly applied in practice, more companies document their operations in the form of process models. Since users require descriptions of one process on various levels of detail, there are often multiple models created for the same process. Business process model abstraction emerged as a technique reducing the number of models to be stored: given a detailed process model, business process model abstraction delivers abstract representations for the same process. A key problem in many abstraction scenarios is the transition from detailed activities in the initial model to coarse-grained activities in the abstract model. This transition is realized by an aggregation operation clustering multiple activities to a single one. So far, humans decide on how to aggregate, which is expensive. This paper presents a semi-automated approach to activity aggregation that reduces the human effort significantly. The approach takes advantage of an activity meronymy relation, i.e., part-of relation defined between activities. The approach is semi-automated, as it proposes sets of meaningful aggregations, while the user still decides. The approach is evaluated by a real-world use case.

1. Meronymy-based
Aggregation of Activities in
Business Process Models
Sergey Smirnov1, Remco Dijkman2,
Jan Mendling3, and Mathias Weske1
1 Hasso Plattner Institute, Germany
2 Eindhoven University of Technology, The Netherlands
3 Humboldt-Universität zu Berlin, Germany

2. Meronymy-based Aggregation of Activities in Business Process Models
2
Meronymy-based Aggregation of Activities in Business Process Models
Motivation
> 300 nodes
> 150 activities

3. Meronymy-based Aggregation of Activities in Business Process Models
3
Business Process Model
Abstraction
… is an operation on a business process model
preserving essential process properties and leaving
out insignificant process details in order to retain
information relevant for a particular purpose

4. Meronymy-based Aggregation of Activities in Business Process Models
4
BPMA Scenario
Abstraction objects = Activities
Abstraction operation = Aggregation
Confirm
order
Revise order
Update
customer
file
Update
manufacturing
plan
Generate
order
summary
Review
order
summary
Arrange
shipment
Receive
order
Update
customer
file
Handle order
Process
order
summary
Receive
order

5. Meronymy-based Aggregation of Activities in Business Process Models
5
P
Structural BPMA Challenges
Confirm
order
Revise order
Update
customer
file
Update
manufacturing
plan
Generate
order
summary
Review
order
summary
Arrange
shipment
Receive
order
Arrange
shipment
Update
manufacturing
plan
Revise order
Generate
order
summary
Review
order
summary
Update
customer
file
Candidate 1
Candidate 2
?Candidate 1 Candidate 2

6. Meronymy-based Aggregation of Activities in Business Process Models
6
Activity Ontology
Create
order
report
Update
customer
profile
Generate
order
summary
Review
order
summary
Process
order
summary
Send
otification
Forward
file to

7. Meronymy-based Aggregation of Activities in Business Process Models
7
P
Role of Activity Ontology in
BPMA
Confirm
order
Revise order
Update
customer
file
Update
manufacturing
plan
Generate
order
summary
Review
order
summary
Arrange
shipment
Receive
order
Generate
order
summary
Review
order
summary
Update
customer
file
Candidate 1
Candidate 2
Create
order
report
Update
customer
profile
Generate
order
summary
Review
order
summary
Process
order
summary
Send
otification
?Candidate 1 Candidate 2

8. Meronymy-based Aggregation of Activities in Business Process Models
8
Aggregation Mining Idea
Input: Process model + Ontology
Output: Aggregations
Algorithm Sketch:
FOR each aggregation candidate
map each aggregation candidate activity to
an ontology activity
IF (ontology activities are strongly related)
aggregation candidate is an aggregation
How to find an aggregation
candidate efficiently?
How to judge on ontology
activity relatedness?

9. Meronymy-based Aggregation of Activities in Business Process Models
9
Activity Alphabet
Create
order
report
Update
customer
profile
Generate
order
summary
Review
order
summary
Process
order
summary
Send
otification
Forward
file to

10. Meronymy-based Aggregation of Activities in Business Process Models
10
Process Model
is a process model, where:
 finite non-empty set of activities
 finite set of gateways

 finite set of nodes
 the flow relation
 a connected graph

11. Meronymy-based Aggregation of Activities in Business Process Models
11
Aggregation Candidate
In process model
is an aggregation candidate.

12. Meronymy-based Aggregation of Activities in Business Process Models
12
Meronymy tree is a tuple
Meronymy Tree
n0
n1 n2 g n3
e f n6 n7 n8n4 n5
n0

13. Meronymy-based Aggregation of Activities in Business Process Models
13
Meronymy Forest
Meronymy forest F is a disjoint union of
meronymy trees
n9
n10 n12
n13
n14 n15 n16
n17 n18 n19 n20 n21
n0
n1 n2 g n3
e f n6 n7 n8n4 n5
n0
n1 n2 g n3
e f n6 n7 n8n4 n5
n9
n10 n12
n13
n14 n15 n16
n17 n18 n19 n20 n21

14. Meronymy-based Aggregation of Activities in Business Process Models
14
Aggregation Candidate
Construction
Construction of aggregation candidates aggregation
through aggregation candidate size increment
start: k =2
i iteration: construct i-size aggregation candidates
from (i-1) aggregation candidates
prune insignificant candidates
prune candidates with large distance
stop: k=|A|, PM = (A, G, E) OR all the aggregation
candidates of size k are pruned

15. Meronymy-based Aggregation of Activities in Business Process Models
15
Activity Match (1)
A1
A2
A3
A4
A5 A6 A7
n0
n1 n2 g n3
e f n6 n7 n8n4 n5
n13
n14 n15 n16
n17 n18 n19 n20 n21
Process model
Meronymy forest

16. Meronymy-based Aggregation of Activities in Business Process Models
16
Activity Match (2)
A1
A2
A3
A4
A5 A6 A7
n0
n1 n2 g n3
e f n6 n7 n8n4 n5
n13
n14 n15 n16
n17 n18 n19 n20 n21
Process model
Meronymy forest

17. Meronymy-based Aggregation of Activities in Business Process Models
17
Activity MixMatch
A1
A2
A3
A4
A5 A6 A7
n0
n1 n2 g n3
e f n6 n7 n8n4 n5
n13
n14 n15 n16
n17 n18 n19 n20 n21
Process model
Meronymy forest n1 g
n14
n15n1
g
n15n14

18. Meronymy-based Aggregation of Activities in Business Process Models
18
n0
n1 n2 g n3
e f n6 n7 n8n4 n5 f
Lowest Common Ancestor
Lowest common ancestor is a function
maps a tree node set to its
lowest common ancestor
e
n2

19. Meronymy-based Aggregation of Activities in Business Process Models
19
n0
n1 n2 g n3
e f n6 n7 n8n4 n5 f
Meronymy Leaves
Meronymy leaves is a function
maps an activity to the leaves
of the subtree rooted to
this activity
e
n2

20. Meronymy-based Aggregation of Activities in Business Process Models
20
n0
n1 n2 g n3
e f n6 n7 n8n4 n5
Degree of Aggregation
Coverage (1)
Degree of aggregation coverage is a function

21. Meronymy-based Aggregation of Activities in Business Process Models
21
n0
n1 n2 g n3
e f n6 n7 n8n4 n5
Degree of Aggregation
Coverage (2)
Degree of aggregation coverage is a function

22. Meronymy-based Aggregation of Activities in Business Process Models
22
n0
n1 n2 g n3
e f n6 n7 n8n4 n5
Degree of Aggregation
Coverage Properties
Shows
 if the LCA has other descendents,
except aggregation candidate
Ignores
 aggregation candidate size
 the aggregation candidate
depth in the LCA subtree
 ignore the LCA depth
Possesses
 value between 0 and 1

23. Meronymy-based Aggregation of Activities in Business Process Models
23
Object Studied in Evaluation
Model collection
 6 process models (42 activities on average)
Meronymy forest
 MIT Process Handbook
 processes elicited in the interviews with process experts
 ≈ 5000 activities
 specifies meronymy and hyponymy
 spans several business domains

24. Meronymy-based Aggregation of Activities in Business Process Models
24
Evaluation Approach
 Each activity aggregation is decomposed into a set
of subsets of size 2, e.g.:
{a, b, c} → {a, b}, {a, c}, {b, c}
 Modeling expert evaluates pair relevance
 Experiments varying node distance, cover
 Observe the precision value

25. Meronymy-based Aggregation of Activities in Business Process Models
25
Evaluation Results
0
0.2
0.4
0.6
0.8
1
1 2 3 4
cover threshold 0.3 cover threshold 0.2
0
0.2
0.4
0.6
0.8
1 2 3 4
cover threshold 0.3 cover threshold 0.2
Observations
0.27 ≤ precision ≤ 0.46
↑ cover → ↑ precision
↑ node distance → ↓ precision

26. Meronymy-based Aggregation of Activities in Business Process Models
26
Conclusion
Contributions
 Metric for relatedness
of activity sets
 Activity aggregation
mining algorithm
Future work
 Improve activity
matching technique
 Precise aggregation
mining technique
evaluation
 Investigate other
information enabling
activity aggregation

27. Meronymy-based Aggregation of Activities in Business Process Models
27
Thank you!

28. Meronymy-based Aggregation of Activities in Business Process Models
28
Contact Details
Sergey Smirnov
PhD Student
Business Process Technology Group
Hasso Plattner Institute
Prof.-Dr.-Helmert-Str. 2-3, 14482 Potsdam, Germany
Email: sergey.smirnov@hpi.uni-potsdam.de
Phone: +49 (0) 331 5509 194
Fax: +49(0) 331 5509 189