Upload
nesma
View
184
Download
0
Embed Size (px)
DESCRIPTION
IWSM Presentation
Citation preview
Automated Functional Size Estimation Using Business Process Models with UPROM Method
Banu Aysolmaz, Onur Demirörs 7 October
Middle East Technical University (METU)Informatics Institute,Software Management Research Grouphttp://smrg.ii.metu.edu.tr
IWSM-MENSURA
BPMod is used to analyze, define and improve business processes (BP)
PAIS is a system developed to automate BPs An organization firstly needs to analyze and define its
business processes to develop PAIS BP knowledge may be utilized in many practices in
SDLC User requirements analysis Functional size estimation Process documentation
Outputs are developed separately and independently, thus end up to be inconsistent and hard to maintain
BPM Context
METU Informatics Institute
FSM methods utilize functional user requirements to identify functional size in FPs
Accurate size measurement is usually available after software requirements specification phase
Project managers need to estimate the effort in earlier phases
Early phase methods are usually unreliable, subjective and unrepeatable
FSM Context
METU Informatics Institute
UPROM: Unified BP Modeling Methodology
Analyze BPs and user requirements in a unified way
Automatically generate artifacts including textual user requirements document, process definition document, business glossary and functional size estimation
UPROM Method
METU Informatics Institute
UPROM Method
METU Informatics Institute
Analyze business
processes and develop BPMod
diagrams
Analyze requiremen
ts and develop analysis diagrams
Generate artifacts
The Process
METU Informatics Institute
STEP 1 STEP 2 STEP 3
Six diagram types are based on common metamodel
The set of models is organized as a modeling project
Objects with the same name and type are assigned to be unique.
The Notation
METU Informatics Institute
Logically related object types are also assigned unique: Information carriers, entity, cluster, attribute, key
attribute, technical term Function, process interface, value chain
METU Informatics Institute
Example FA Diagram
METU Informatics Institute
• carries out• approves• supports• contributes to• must be
informed on completion
• uses • views • creates• changes• reads• Deletes• lists
Example (Partial) Conceptual ER Diagram
METU Informatics Institute
1. Measurement Strategy Phase Software in COSMIC => Application Functional users => Organizational elements
External applications 2. Mapping Phase
Functional Process => Functions in EPC (to be automated)
Triggering event => Event before the related function
OOI => Entities on FADs DG => Implicit (usually indicated in the constraint)
UPROM FSE Method – Apply Phase 1 & 2
METU Informatics Institute
3. The Measurement Phase For each FAD, Convert «Operations» to DMs
Then apply 6 Rules (4 on individual FADs, 2 on the whole modeling project)
UPROM FSE Method – Apply Phase 3 Conversion
METU Informatics Institute
Operation Type
Base CRUDL Operation
Data Movements
Create create E, W
Change update, list E, R, W, X
Delete delete E, W
View list E, R, X
List list E, R, XRead read R
Use read R
Payments Web Service: 1 creates + 1 uses = 1E + 1R + 1W
eCompany Application Module: 5 Views + 1 Create + 1 Change = 7E + 6R + 5W + 6XMETU Informatics Institute
Rule 1: Add X DMs if missing«If there is a Create or Delete operation but no X DM is introduced, add one X DM»
UPROM FSE Method – Apply Phase 3 Rule 1
METU Informatics Institute
Application:
Add 1 X to Payments Web Service
Payments Web Service:
1E + 1R + 1W + 1X
Application:
Remove one E DM as «external approvals» and «cabinet approval» are related.
eCompany Application Module: 6E + 6R + 5W + 6X
Rule 2: Remove extra E DMs for «Related Entities»
E’s may be accumulated unnecessarily due to multiple operations depicted on related entities
UPROM FSE Method – Apply Phase 3 Rule 2
METU Informatics Institute
METU Informatics Institute
Rule 3: Multiply DMs for multiple functional users
UPROM FSE Method – Apply Phase 3 Rule 3
METU Informatics Institute
Application:
Multiply DMs for «carries out» and «approves» roles
eCompany Application Module: 12E + 12R + 10W + 12X Payments Web Service 2E + 2R + 2W + 2X
Rule 4: At least one operation of type create, change, delete, view and list must exist
, a functional process shall comprise at least two DMs, an E plus either an X or a W
UPROM FSE Method – Apply Phase 3 Rule 4
METU Informatics Institute
Rule 5: Add DMs of external applications to the main application on that FAD
UPROM FSE Method – Apply Phase 3 Rule 5
METU Informatics Institute
Application:
Add «Payments Web Service» DMs to the other
eCompany Application Module: 14E + 14R + 12W + 14X
Rule 6: An Entity must be «created» in at least one FP!
UPROM FSE Method – Apply Phase 3 Rule 6
METU Informatics Institute
Check: “external approvals, cabinet approval, license, land register, bank guarantee, application status” entities are created!
Report results for EPCs and FADs
Sum up total FPs for applications
UPROM FSE Method – Aggregate Results
METU Informatics Institute
Celebrity System
Veterinary System
Movie Manager
# EPC & FT diagrams 5 4 3
# FA diagrams 8 7 11# entities 4 4 6
Case Study 1 – Three Simple Business Applications
METU Informatics Institute
BPs are modeled, FSE generated and compared with existing FSM results.
Celebrity System
Veterinary System
Movie Manager
Average
COSMIC FSM Size 36 CFP 37 CFP 85 CFP 53 CFP
UPROM FSE Size 38 CFP 36 CFP 83 CFP 51 CFP
Deviation of UPROM FSE size
5,6% -2,7% -2,4% 3,6%
10% sample is selected in BPs and manually measured by an expert
Results compared with UPROM FSE results
Case Study 2 – Two e-Government Projects
METU Informatics Institute
Entry Read Write Exit TotaleCompany
332 290 200 296 1118 CFP30% 26% 18% 26%
eTrademark119 112 89 117 437 CFP27% 26% 20% 27%
Deviation for absolute error of each FP 24%Deviation in total size 8%
FSE is conducted automatically, virtually no extra effort is required
«Reasonably accurate» FSE is achieved in an early phase
Variation and subjectivity due to the measurer decreased
Measurement can be repeated easily upon requirements update
No specific training or expertise required for COSMIC/FSM
Conclusion
METU Informatics Institute
Further resources: B. Aysolmaz and O. Demirors, “Modeling Business Processes to
Generate Artifacts for Software Development : A Methodology,” in Modeling in Software Engineering (MISE), 2014 ICSE Workshop on, 2014.
B. Aysolmaz, “Unified Business Process Modeling Methodology (UPROM) Application: Case Study Results METU/II-TR-2014-32,” Ankara, Turkey, 2014. http://expertjudgment.com/publications/METU_II_TR_2014_32.pdf
B. Aysolmaz, “UPROM: A Unified Business Process Modeling Methodology,” Middle East Technical University, 2014.
B. Aysolmaz and O. Demirörs, “UPROM Tool : A Unified Business Process Modeling Tool for Generating Software Life Cycle Artifacts,” in Proceedings of the CAiSE Forum 2014, 2014.
B. Aysolmaz and O. Demirors, “Deriving User Requirements from Business Process Models for Automation: A Case Study,” in Requirements Engineering and Business Process Management Workshop, RE'14, 2014.
Thank you for listening…
METU Informatics Institute