Improving Business Processes using Process-oriented Data Warehouse
Muhammad Khurram Shahzad
Doctoral Dissertation in
Computer and Systems Sciences
Supervised by: Paul Johannesson Jelena Zdravkovic
Agenda
• Introduction
• Research Question and Research Goal
• Research Methodology
• The Proposed Artifacts
• Evaluation
• Conclusion
Introduction
• The BPM lifecycle consists of four phases, process design, process implementation, process enactment and performance evaluation [1, 2]• “The traces stored in logs are widely acknowledged as significant for analyzing performance of processes to identify opportunities for improvement” [3, 4]
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Introduction
• However, execution logs cannot be used [4, 5, 6], because- Logs capture traces for short time- During process execution, logs are continuously
updated- Data from other sources cannot be added to
process logs due to their design limitations• Solution: Data warehousing and data mining [4, 5, 7]
data warehousing
and data mining
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Why Data Warehousing?
• According to DM review – a premier magazine on BI• The market of business intelligence tools and
techniques raised to 13.4 billion in 2003
• According to the 451 Research*
• Among these, specifically data warehousing market has seen fastest growth*
• Annual growth rate from 2009 is 11.5%, and it is projected to be 13.2 billion dollar in revenue by 2013*
*also a consortium of companiesProblem
Awareness
Suggestion &
Development
Evaluation
Conclusion
Process Warehouse vs. Data Warehouse
• Process Warehouse (PW) is a specialized data warehouse used for performance analysis and improvement of processes
“PW provides comprehensive information on processes quickly, at various aggregation levels and from multidimensional points of view” [6]
• PW differs from data warehouse because it designed to store process traces
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Problem Space
• PW is a large, and the magnitude of data needed for process performance analysis and decision making is small compared to PW size• Selection of appropriate dimensions may require significant domain expertise• Higher cognitive effort to extract and interpret the information from PW will not bring any value to the decision maker
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Research Question and Goal
• How to facilitate performance analysis and improvement of business processes using
process warehouse?
Goal - To develop a method for performance analysis of processes and deciding on process
improvements using process warehouse.
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Research Approach
• IS research is classified into two research paradigms [8, 9]• Behavioral Science – justifying theories to explain
human and organizational behavior• Design Science – problem solving paradigm to
create (technology oriented) artifacts [9]
•We use Design Science
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Research Approach
•Design Science – problem solving paradigm to create technology-oriented artifacts [9]
Phases of design science [10]
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Suggestion and Development
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Suggestion and Development
•Our approach is based on integration of goals with PW• To allow goal-based navigation of PW•We propose • A Process Warehouse• A method for using PW for process analysis and
improvement
Quality of service state of a process intended to be achieved. Like, efficient, timely, safe*
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
*Swedish Institute of Medicine
Recall, PW is large, navigation require expertise, higher cognitive effort
The Proposed Process Warehouse
•Our PW differs from a PW in a number of ways that spans across two levels, - Structural level describes the design
specification of data, relationship between data and constraints in a data
- Architectural level is the set of specifications that describes the organization of warehouse objects, how they work together and how the data flows between them
✔
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Process Warehouse: Structural level
• At structural level our PW differs from a PW, because it consists of two parts, stable and case specific - The stable part, to captures information about goals,
indicators, satisfaction conditions and their relation with PW• This part is hard coded
- The case specific part, captures the dimensions and facts essential for performance analysis of processes • This part is changeable (dynamic)
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Process Warehouse: Architectural level
• For populating the case-specific part of PW, data needs to be extracted and consolidated from process logs as well as from the transactional sources, which is not the case with traditional PW
Process Warehouse
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
The Proposed Method
Step 1:
•Build Goal structure
Step 2
•Integrate Goals with Process Warehouse
Step 3
•Performance Analysis and Improvement
✔
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
The Method – Step 1
Step 1• Build Goal structure
•Recursively analyze Business Process
Task 1
•Identify goals of the Process & decompose
Task 2
•Identify criteria for fulfillment of goals
Task 3
Process Decomposition Tree
✔
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Modular decomposition of the control structure of a process
Goal Decomposition Tree
The Method – Step 1
Step 1• Build Goal structure
•Recursively analyze Business Process
Task 1
•Identify goals of the Process & decompose
Task 2
•Identify criteria for fulfillment of goals
Task 3
Goal Decomposition Tree
✔
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Hierarchical structure of goals aligned with modular decomposition of a process
Output: Goal Decomposition Tree
The Method – Step 2
Step 2
• Integrating Goals with Process Warehouse
•Concepts needed to relate goals with PW
Conceptual level
•Extensions to PW design specification to integrate goals
Implementation level
✔
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Output: Goal –PW Integration
The Method – Step 2
Step 2
• Integrating Goals with Process Warehouse
•Concepts needed to relate goals with PW
Conceptual level
✔
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Bitmap attribute
Bitmap attribute
Process Warehouse
The Method – Step 2
Step 2
• Integrating Goals with Process Warehouse
•Extensions to PW design specification to integrate goals
Implementation level
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Stable part of PW
Case-specific part of PW
✔
The Method – Step 3
Step 3• Analyze and Improve Process
•Condition Identification
Task 1
•Goal Identification
Task 2
•Information Analysis
Task 3
Task 4
Task 5
• Decision Elicitation
• Process Change Solution
✔
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
The Method – Step 3
Step 3• Analyze and Improve Process
•Condition Identification
Task 1
•Goal Identification
Task 2
•Information Analysis
Task 3
Task 4
Task 5
• Decision Elicitation
• Process Change Solution
Navigation Operations
✔
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Traverse down, traverse up, traverse across, iterative traverse across
The Method – Step 3
Step 3• Analyze and Improve Process
•Condition Identification
Task 1
•Goal Identification
Task 2
•Information Analysis
Task 3
Task 4
Task 5
• Decision Elicitation
• Process Change Solution
Suitability Estimation Model
✔
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Type level – suitability function µInstance level – convenience σ
Evaluation
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Evaluation
• March [9] suggested two sequential steps for evaluation for design science- Criteria development- Assessment of artifact against the criteria
• We use Moody’s Method evaluation model [11] , because- It is widely used for evaluation of IS artifacts- It incorporates performance and perception based
evaluation• For perception based evaluation we adopt the
evaluation model of Hong’s model [12] because – It is based on Technology acceptance model and IS
success model– Also consider factors affecting DW success
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Evaluation
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
• In addition to that, mandatory elements of the method [12]
Prototype
Introduction
Research Question
Contribution
Conclusion
Performance based Evaluation
• Accessible facts remains fixed with traditional approach, but changes with our goal based approach
• The cognitive efforts to interpret information is reduced
Accessible facts
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Performance based Evaluation
• Accessible dimensions remains fixed with traditional approach, but changes with our goal based approach
• The domain expertise required to select appropriate dimension
Accessible dimensions
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Performance based Evaluation
• Increase in precision affirms the retrieval of relevant data
Comparison of precision
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Perception based Evaluation
• The method overall received a positive response• This indicates that the method was found to be useful
Frequency distribution of constructs
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
PEOU - Perceived easy of usePU – Perceived usefulness
improved task outcomeImprove analysis performanceHelp making better decisions
Easy to learnEasy to get required infoEase to become expert user
Help finishing task quicklyUseful for analysisHelp improving analysis taskEasy to locate dataEasy to use data access toolsSufficient data access tools
CompletenessGranularitySufficiency
Perception based Evaluation
• Experienced users agreed in larger percentage than novice
• Indicates construct items are better perceived by experience users than novice users
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Frequency distribution of constructs
Sufficient training
Conclusion
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Conclusions
• The method provides a step by step approach that can facilitate process analysis and improvement
• Results indicate that use of the proposed method has been perceived positively
• Due to traceability between goals and PW content, relevant content is retrieved
• Due to goal based navigation the task of navigating through PW is simplified
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
Acknowldgements
Problem Awarenes
s
Suggestion &
Development
Evaluation
Conclusion
References
[1] M. Weske, W.M.P. van der Aalst, H.M.W. Verbeek. Advances in business process management. Data and Knowledge Engineering, 50(1), pp. 1-8, 2004.
[2] M. zur Muhlen. Workflow-based process controlling: Foundations, Design, and Application of Workflow-driven Process Information Systems. 1st edition, Logos Verlag Berlin, 2004.
[3] W. van der Aalst, Mariska Netjes and Hajo A. Reijers. "Supporting the Full BPM Life-Cycle Using Process Mining and Intelligent Redesign."Contemporary Issues in Database Design and Information Systems Development. IGI Global, 2007. 100-132. Web. 13 Dec. 2011. doi:10.4018/978-1-59904-289-3.ch004.
[4] D Grigori, F Casati, M Castellanos, U Dayal, M Sayal, M C Shan. Business Process Intelligence. Computer in Industry 53(4), pp. 321-343, 2004.
[5] M Castellanos, A Simitsis, K Wilkinson, U Dayal. Automating the loading of business process data warehouses. Proceedings of the 12th International Conference on Extending database technology: Advances in Database Technology (EDBT'09), Russia.
References
[6] B. List, J. Schiefer, A.M. Tjoa, G. Quirchmayr. Multidimensional business process analysis with the process warehouse. Knowledge discovery for business information systems, Vol 600, pp. 211-227, Kluwer Publications, 2002.
[7] T. Bucher, A Gericke. Process-centric business intelligence. Business Process Management Journal, 15(3), pp. 408-429, 2009.
[8] A.R. Hevner, S.T. March, J. Park. Design Science in Information Systems Research, MIS Quarterly, 28 (1), pp. 75-105, 2004.
[9] S.T. March, G.F. Smith. Design and natural science research on information technology. Decision Support Systems, 15 (4), 251-266, 1995.
[10] H. Takeda, P. Veerkamp, T. Tomiyama, H. Yoshikawam. "Modeling Design Processes." AI MagazineWinter: 37-48, (1990).
[11] Daniel L. Moody: The method evaluation model: a theoretical model for validating information systems design methods. In proceedings of the European Conference on Information Systems (ECIS'2003), pp. 1327-1336, Italy.