Andres jimenez c ai-se13 presentation

Generating Multi-objective Optimized
Business Process Enactment Plans
25th International Conference on
Advanced Information Systems Engineering
2013
Andrés Jiménez, Irene Barba, Carmelo del Valle and Barbara Weber
Departamento de Lenguajes y Sistemas Informáticos. University of Seville, Spain
{ajramirez, irenebr, carmelo}@us.es
Department of Computer Science, University of Innsbruck, Austria
barbara.weber@uibk.ac.at

CAiSE 2013 – 17-21 June, Valencia (Spain) 2/33
System
Configuration
Process
Enactment
Evaluation
Process
Design &
Analysis
BPM lifecycle

Designing the model
Ferreira, H.M. et al. (2006)
Karim, A. et al. (2013)

Flexible design

CAiSE 2013 – 17-21 June, Valencia (Spain)
A declarative language for modelling dynamic business processes
1) Tasks (smallest
unit of work)
2) Relations (constraints,
no order of execution)
A B C
0..2 1
if A is executed, B
is executed and
vice versa
B can be executed
at most twice
every B is
eventually
followed by C
C is executed
once
Declare (2006)
Declarative languages
Pesic, M. and van der Aalst, W.M.P. :
(2006)
5/33

Just say what, and
let the AI tell you
how.
Our proposal
6/33

Just say what, and
let the AI tell you
how.
Our proposal
7/33

Just say what, and
let the AI tell you
how.
Our proposal
8/33

Recommendations
Just say what, and
let the AI tell you
how.
Our proposal
9/33

Outline
1. Background & Introduction
2. The What. Extension of Declare
3. The How. BP Enactment Plans
4. Constraint Satisfaction Problems and Optimization
5. Future work
10/33

2. Declare-R an extension of Declare
Estimates + Resources + Multiple Instances + Data + Temporal
(0, 10)
Client Data (client)
{clientName, bookedServic
es, appointmentTime}
this.startTime ≥ client.appointmentTime
20
Different activity
attributes
11/33

Services

3. Enactment Plans how is it executed
R1
A
0..2
4
1
3
2
R1
C
1
1
1 Res. Availability
#R1: 1
#R2: 2
profit
duration
R2
B
14/33

Plan 1
t = 0 1 2 3 4
R1
R2
A A A A C
B B B
Res. Availability
#R1: 1
#R2: 2
15/33
profit
duration
R1
A
0..2
4
1
3
2
R1
C
1
1
1
R2
B

Plan 1
t = 0 1 2 3 4
R1 A A A A C
R2 B B B
t = 0 1 2 3 4 5 6
R1
R2
Plan 2
A A A A C
B B B B B B
Res. Availability
#R1: 1
#R2: 2
16/33
profit
duration
R1
A
0..2
4
1
3
2
R1
C
1
1
1
R2
B

Plan 1
t = 0 1 2 3 4
R1 A A A A C
R2 B B B
t = 0 1 2 3 4 5 6
R1 A A A A C
R2 B B B B B B
Plan 2 Plan 3
t = 0 1 2 3 4
R1
R21
R22
A A A A C
B B B
B B B
Res. Availability
#R1: 1
#R2: 2
17/33
profit
duration
R1
A
0..2
4
1
3
2
R1
C
1
1
1
R2
B

Plan 1
t = 0 1 2 3 4
R1 A A A A C
R2 B B B
t = 0 1 2 3 4 5 6
R1 A A A A C
R2 B B B B B B
Plan 2 Plan 3
t = 0 1 2 3 4
R1
R21
R22
A A A A C
B B B
B B B
Plan 4
t = 0
R1 C
Total time: 5
Total profit: 4
Total time: 7
Total profit: 6
Total time: 5
Total profit: 6
Total time: 1
Total profit: 1
Minimize total time
Maximize total profit
Res. Availability
#R1: 1
#R2: 2
18/33
profit
duration
R1
A
0..2
4
1
3
2
R1
C
1
1
1
R2
B

Plan 1
t = 0 1 2 3 4
R1 A A A A C
R2 B B B
t = 0 1 2 3 4 5 6
R1 A A A A C
R2 B B B B B B
Plan 2 Plan 3
t = 0 1 2 3 4
R1
R21
R22
A A A A C
B B B
B B B
Plan 4
t = 0
R1 C
Total time: 5
Total profit: 4
Total time: 7
Total profit: 6
Total time: 5
Total profit: 6
Total time: 1
Total profit: 1
Minimize total time
Maximize total profit
Res. Availability
#R1: 1
#R2: 2
19/33
profit
duration
R1
A
0..2
4
1
3
2
R1
C
1
1
1
R2
B

4. Constraint Satisfaction Problem
A CSP is composed by
- a set of variables,
- a domain of values for each variable,
- and a set of constraints between variables.
20/33
The solutions of a CSP are all the possible
combinations of values of the variables which
satisfy the constraints.
search algorithm

4. Constraint Satisfaction Problem
Solve a Constraint Satisfaction /
(CSP/COP)
Generate an
Enactment Plan Optimization Problem
Res. Availability
#R1: 1
#R2: 2
Number of times the
activity is executed
resource selection
High level
constraints
Optimization
Minimize(OCT)
Overall
completion
time
21/33
R1
A
0..2
1
4
2
3
R1
C
1
1
1
R2
B
Start time

OF2
OF1
4. Multi-objective approach
22/33

OF2
OF1
23/33
Ɛ-constraint method

OF2
OF1
24/33
Ɛ-constraint method

OF2
OF1
Pareto Front solutions
25/33

Low work load
High work load
Number of
clients
Waiting Time
or Profit
15 minutes of
waiting time!

Future Work
- Robustness
t = 0 1 2 3 4 5 6 7
R1 A1 A2 A2 A2 A2 A2 C2
R21 B2 B2 B2
R22 B2 B2 B2
t = 0 1 2 3 4 5 6 7
R1 A1 A2 A2 A2 A2 A2 C2
R21 B2 B2 B2 B2 B2 B2
Same completion time
Same total profit
- Stochastic attributes
R1
C
[1..5]
1 27/33

Thank you
Any question?
21st International Conference on
Information Systems Development
2012
Andrés Jiménez Ramírez
Departamento de Lenguajes y Sistemas Informáticos.
University of Seville, Spain
ajramirez@us.es

Applications
1) Simulation
2) Time prediction
3) Recommendations
4) Generation BP models
29/33

1) Simulation
2) Time prediction
3) Recommendations
30/33
Applications

1) Simulation
2) Time prediction
3) Recommendations
What-if scenarios
(reduce resources
change estimates,
etc.)
31/33
Applications

1) Simulation
2) Time prediction
3) Recommendations
What-if scenarios
(reduce resources
change estimates,
etc.)
32/33
Applications

1) Simulation
2) Time prediction
3) Recommendations
Predicting the
completion time of the
running instances
33/33
Applications

1) Simulation
2) Time prediction
3) Recommendations
Predicting the
completion time of the
running instances
34/33
Applications

1) Simulation
2) Time prediction
3) Recommendations
Partial
traces
35/33
Applications

1) Simulation
2) Time prediction
3) Recommendations
Partial
traces
36/33
Applications

1) Simulation
2) Time prediction
3) Recommendations
Convert enactment plans to
BP models in standard BPMN
A B C
0..2 1
R1
4
R2
3
R1
1
A C
+
B1
B2
R
1
R
2
Plan
37/33
Applications

Andres jimenez c ai-se13 presentation

More Related Content

Similar to Andres jimenez c ai-se13 presentation (20)

More from caise2013vlc (20)

Recently uploaded (20)

Andres jimenez c ai-se13 presentation