Corporate Profile 47Billion Information Technology
Scekic caise13-
1. 25th International Conference on Advanced Information Systems Engineering
June 17-21, 2013, Valencia, Spain
Ognjen Scekic, Hong-Linh Truong, Schahram Dustdar
Distributed Systems Group
Vienna University of Technology
http://dsg.tuwien.ac.at
Programming Incentives
in Information Systems
2. 2 CAiSE’13
Evolution of Collaborative Processes
Conventional workflows
• formal description
• structured execution
• predefined roles and activities
• complex tasks
Crowdsourcing
• simple tasks
• anonymous replaceable actors
• short, unstructured interactions
• No interaction/collaboration
among actors
+
=
Socio-technical Collective Adaptive System
• ad-hoc assembled teams
• complex tasks
• social orchestration
• indirect adaptation
3. 3 CAiSE’13
Programmable incentive
management
Requirements:
– Modeling
– Programming
– Execution
– Monitoring
– Re-use
Incentive Programming Model for CASs
EU FP7 SmartSociety project
www.smart-society-project.eu
4. 4 CAiSE’13
Incentives & Rewards
• Incentives
Stimulate (motivate) or discourage
certain worker activities before the
actual execution of those activities.
• Rewards
Any kind of recompense for worthy
services rendered or retribution for
wrongdoing exerted upon workers
after the completion of activity.
• Incentive Mechanism
A plan (rule) for assigning rewards.
5. 5 CAiSE’13
We identified 7 basic
incentive mechanisms in use
today and their constituent
elements.
New mechanisms can be built
by composing and
customizing well-known
incentive elements.
Portable, reusable, scalable
Modeling Incentives
7. 7 CAiSE’13
PRogrammable INCentives Framework (PRINC)
Representation of external system suitable for modeling application of
incentives.
• State – Global state, individual worker attributes and performance metrics
(QoS).
• Time – Records of past and future worker interactions supporting time
conditions.
• Structure – Representation and manipulation of various types of relationships
Rewarding
Model
(RMod)
8. 8 CAiSE’13
Examples of mechanisms that RMod can encode and
execute:
− At the end of iteration, award each worker who scored better
than the average score of his immediate neighbors.
− Unless the productivity increases to a level p
within n next iterations, replace team's current manager
with the most-trusted of his subordinate workers.
The Rewarding Model (RMod)
9. 9 CAiSE’13
PRINC Framework
• Definition of system-specific artifacts, actions, attributes and
relation types.
• Definition and parameterization of
metrics, messages, structural patterns and custom incentive
mechanisms.
Mapping
Model
(MMod)
10. 10 CAiSE’13
The Mapping Model (MMod)
Example: Adapting a general incentive mechanism for a software testing
company.
DSL
When a bug
report is
verified, award
points to the
submitter.
library
11. 11 CAiSE’13
PRINC Framework
• Declarative, domain-specific language.
• High-level, platform independent, human-friendly
notation.
Incentive
Model
(IMod)
12. 12 CAiSE’13
We do not invent nor evaluate incentive mechanisms.
Basic techniques, such as composition of mechanisms
evaluated through simulation:
DomainPro1 tool
Evaluation
1 http://quandarypeak.com/
15. 15 CAiSE’13
Conclusions:
– Socio-technical systems need effective incentive management.
– We presented a framework for
modeling, composing, adapting, executing and monitoring
portable incentive strategies.
Current work:
– High-level, user-friendly, graphical DSL.
– Integration into the overall programming model for CASs.
Future Work:
– Determine best incentive practices in a given environment
by learning from past incentive applications.
Conclusion & Future Work
16. 25th International Conference on Advanced Information Systems Engineering
June 17-21, 2013, Valencia, Spain
Ognjen Scekic, Hong-Linh Truong, Schahram Dustdar
Distributed Systems Group
Vienna University of Technology
http://dsg.tuwien.ac.at
Modeling Rewards and Incentive Mechanisms for Social BPM
Thank you!
Questions?
Editor's Notes
Information systems managing humans as first class actors.
At the heart of the system lies RMod.Abstraction layer between an actual platform that manages worker teams and client's system-independent representationof an incentive mechanism. General enoughto model many different, real-world platforms and support most incentive mechanisms.Low-level andminimalistic.
Customizing incentives and adapting them to the underlying system.
Similar piece-work tasks: tagging images, translating text.