Integrating Process Learning and ProcessEvolution − A Semantics Based Approach

Stefanie Rinderle, Barbara Weber,Manfred Reichert, Werner Wild

Nancy, September 6th 2005

International Conference on Business Process Management

Outline

1. Motivation

2. The Process Life Cycle

3. Providing Process Change Semantics through CCBR

4. Process Learning and Seamless Process Evolution

5. Case-Base Evolution

6. Summary and Outlook

Appendix: References

Motivation

Situation in Practice (1)

υ How quickly can new processes be implemented?υ At which costs?υ How expensive are process changes in the sequel?υ How to avoid the maintenance trap?

Enterprises• external services• internal processes

E-Commerce

M-Commerce

B2B

CRM

E-Procurement

Web-ShopsElectronic

Market Places

Permanently arising new „trends"

– require new products and services

Real-TimeEnterprise

questions:

... which must be integrated

Challenges!

WebServices

– Change risk?

4©Stefanie Rinderle, Dept. DBIS, University of Ulm, 2005

π Processes with long duration

❍ leasing contracts (3 – 5 years)❍ medical treatments (up to several months)❍ ...

π Consequence: processes have to be frequently adapted

❍ new laws❍ new medical treatment❍ operational reorganization❍ …

Motivation

Situation in Practice (2)

5©Stefanie Rinderle, Dept. DBIS, University of Ulm, 2005

Adaptive Process Management

Instance I1:

Instance I2:

Instance I3:

Instance I1 on S‘Migration

unbiased

Migration Instance I2 on S‘ (ad hoc modified)

disjointsend form send shirt

move

confirm order pack goods

delivergoods

getorder

delivergoods

sendform

getorder

packgoods

confirmorder

S‘:sendshirt

send form send shirt

move

S:

Motivation

overlapping

Instance I3 on S‘Migration

Complete Framework at Syntactical Level

6©Stefanie Rinderle, Dept. DBIS, University of Ulm, 2005

Outline

1. Motivation

2. The Process Life Cycle

3. Providing Process Change Semantics through CCBR

4. Process Learning and Seamless Process Evolution

5. Case-Base Evolution

6. Summary and Outlook

Appendix: References

7©Stefanie Rinderle, Dept. DBIS, University of Ulm, 2005

Challenges

Semantic Process Change Support:

π Explaining / Documenting Process Instance Changes

π Reviewing Reasons for Previous Process Instance Changes

π Re-using Previous Process Instance Changes

π Deriving Suggestions for Process (Type) Optimizations

π Automatic Process Instance Migration

Process Life Cycle

Combining Adaptive PMS with Case-Based Reasoning

The Big PictureProcess Life Cycle

Lab

test

Add / Reuse

Case LabTest

I! (! = 1...n)

I!

(! = 1...n)

Changed Process

Instances

Lab

test

CCBR

Inst

antia

tion P

rocess

Typ

e C

hange

Process Instance Change

Notification

Threshold exceeded

Process Instances

Process User

Process Engineer

Process

Engineer

Migrate

case-base

Prepare

Patient

Examine

patient

Make

appointment

Schema S‘:

Enter

order Inform

patient

Lab

Test

Make

appointment

Deliver

report

Prepare

Patient

Schema

S:

Enter

order Inform

patient

Prepare

Patient

Examine

patient

Deliver

report

Make

appointment

Lab

test

Add / Reuse

Case LabTest

I! (! = 1...n)

I!

(! = 1...n)

Changed Process

Instances

Lab

test

CCBR

Inst

antia

tion P

rocess

Typ

e C

hange

Process Instance Change

Notification

Threshold exceeded

Process Instances

Process UserProcess User

Process EngineerProcess Engineer

Process

Engineer

Process

Engineer

Migrate

case-base

Prepare

Patient

Examine

patient

Make

appointment

Schema S‘:

Enter

order Inform

patient

Lab

Test

Make

appointment

Deliver

report

Prepare

Patient

Schema

S:

Enter

order Inform

patient

Prepare

Patient

Examine

patient

Deliver

report

Make

appointment

9©Stefanie Rinderle, Dept. DBIS, University of Ulm, 2005

Outline

1. Motivation

2. The Process Life Cycle

3. Providing Process Change Semantics through CCBR

4. Process Learning and Seamless Process Evolution

5. Case-Base Evolution

6. Summary and Outlook

Appendix: References

10©Stefanie Rinderle, Dept. DBIS, University of Ulm, 2005

Enriching Process Instance Changes Using CCBRProviding Change Semantics

Workflow User

ProcessInstance I:

Enterorder

Informpatient

PreparePatient

Examinepatient

Deliverreport

Makeappointment

Labtest

Insert (LabTest, Prepare Patient, Examine Patient)

Case-Base

Add Case

Add CaseTitle

Description

Question-Answer Pairs

Question AnswerPatient has diabetes? Yes

What is the patient’s age? > 40

Actions

sInsert LabTest S, PreparePatient, ExaminePatient

Operation Type Subject Parameters

Retrieve Case

SelectOperation Type Insert

SelectActivity/Edge Lab Test

PleaseAnswer the Questions

Question AnswerPatient has diabetes? Yes

What is the patient’s age? > 40

Lab Test requiredTitle

125CaseID

100%Similarity

25Reputation Score

Display List of Cases

11©Stefanie Rinderle, Dept. DBIS, University of Ulm, 2005

Outline

1. Motivation

2. The Process Life Cycle

3. Providing Process Change Semantics through CCBR

4. Process Learning and Seamless Process Evolution

5. Case-Base Evolution

6. Summary and Outlook

Appendix: References

Performing Process Type Changes (1)Process Learning/Evolution

Different Scenarios:

• Equivalent cases

• Same actions but different reasons

• Partially overlapping actions

Performing Process Type Changes (2)Process Learning/Evolution

serial insert at process instance level

conditional insert atprocess type level

Performing Process Type ChangesProcess Learning/Evolution

Process Type Level:

Enterorder

Examinepatient

Deliverreport

Makeappointment

PreparePatient

Schema Version S := S(T,1)

Enterorder

Lab test

Examinepatient

Makeappointment

sc1:age > 40 ∧diabetes=„yes“

PreparePatient

patData patData

sc2: default

Schema Version S‘ := S(T,2)

I1 on S:

CompletedActivated CompletedActivated

Process Instance Level:

Lab test

Lab test

ΔT = cInsert(S, Lab test, Prepare Patient,

Examine Patient, sc1)

Migration Policy1: adapt markings

Migration Policy 1: adapt markings

I1 on S‘:unbiased

I3 on S:

I2 on S: disjointbias Migration Policy 2: adapt markings + keepbias on S‘

Migration Policy 2: adapt markings+ keepbias on S‘

I2 on S‘:

subsumption equivalentbias

I4 on S:

ΔI3(S)= {sInsert(S , Lab test, Prepare Patient, Examine Patient)}

I3 on S‘:Migration Policy 3: adapt markings+

bias on S‘ = ∅

Migration Policy 3: adapt markings +

bias on S‘ = ∅Δ

I3(S‘)= ∅

ΔI4(S)= {sInsert(S , Lab test, Prepare Patient, Examine Patient),

delAct(S , deliver Report)

I4 on S‘:

ΔI4(S‘)= [delAct(S ‘, deliver Report)}

prov idesuggestionto user

prov idesuggestionto user

partiallyequivalent

Deliverreport

ΔI2(S)= {delAct(S , deliver Report)}

15©Stefanie Rinderle, Dept. DBIS, University of Ulm, 2005

Outline

1. Motivation

2. The Process Life Cycle

3. Providing Change Semantics through CCBR

4. Process Learning and Seamless Process Evolution

5. Case-Base Evolution

6. Summary and Outlook

Appendix: References

Case-Base MigrationCase-Base Evolution

CCBR:

ΔT1 = {sInsert(S, X, C, E), deleteAct(S, D)} ΔT2 = {deleteAct(S‘, F)}

Process Type Level:

Enterorder

Examinepatient

Deliverreport

Makeappointment

A C D EPreparePatient

B

Schema Version S := S(T,1)

Enterorder

Lab test

Examinepatient

Deliverreport

Makeappointment

A C

X

D E

sc1: age > 40 ∧diabetes =„yes “

PreparePatient

B

patData patData

sc2: default

Schema Version S‘ := S(T,2)

ProcessType Change ΔT

cb := cbT,1:

c1: (..., {sInsert(S, X, C, D)}), f req S(c1) = 48c2: (..., {sInsert(S, X, C, D)}), f req S(c2) = 23c3: (..., {sInsert(S, X, C, D)}), f req S(c3) = 33c4: (..., {sInsert(S, Y, A, B)}), f reqS(c4) = 5c5: (..., {deleteAct(S, D)}), freqS(c5) = 60c6: (..., {deleteAct(S, E),

sInsert(S, Y, A, B)}), f reqS(c6) = 2

cb‘ := cbT,2:

c4: (..., {sInsert(S, Y, A, B)}), c5: (..., {deleteAct(S, D)})c6: (..., {deleteAct(S, E),

sInsert(S, X, C, D)})

c7: (..., {sInsert(S, Y, A, B)})c8: (..., {deleteAct(S, B)})

c1, c2, c3 c4, c6c5c7, c8

dropped by process engineerautomatically transferredtransfered by process engineernew cases for instances based on S'

Migrationc1, c2, c3 c4, c6c5c7, c8

dropped by process engineerautomatically transferredtransfered by process engineernew cases for instances based on S'

Migration

ΔT = cInsert(S, Lab test, PreparePatient, ExaminePatient, sc1)

17©Stefanie Rinderle, Dept. DBIS, University of Ulm, 2005

Outline

1. Motivation

2. The Process Life Cycle

3. Providing Change Semantics through CCBR

4. Process Learning and Seamless Process Evolution

5. Case-Base Evolution

6. Summary and Outlook

Appendix: References

18©Stefanie Rinderle, Dept. DBIS, University of Ulm, 2005

Summary and Outlook

π Combining Adaptive Process Management and CCBR results in

µ Full Life Cycle Support

µ Explanation of Process Instance Changes

µ Derivation of Process Type Changes

µ Process Instance and Case-Base Migration

π Future Work:

µ Implementation

µ Combination with Process Mining Techniques

Summary and Outlook

THANK YOU FOR YOUR ATTENTION!

Own Publications:

S. Rinderle, M. Reichert, and P. Dadam. Supporting Workflow Schema Evolution by Efficient ComplianceChecks and Automatic Migration of Workflow Instances. Informatik - Forschung und Entwicklung,17(4):177–197, 2002. (in German)

S. Rinderle and P. Dadam. Schema Evolution in Workflow Management Systems. Informatik Spektrum,26(1):17–19, 2003. (in German)

S. Rinderle, M. Reichert, and P. Dadam. Evaluation of Correctness Criteria for Dynamic WorkflowChanges. In Proc. Int’l Conf. BPM’03, pp. 41–57, Eindhoven, The Netherlands, 2003

M. Reichert, S. Rinderle, and P. Dadam. On the Common Support of Workflow Type and InstanceChanges under Correctness Constraints. In Proc. Int’l Conf. CoopIS’03, pp. 407–425, Catania, Italy,November 2003

S. Rinderle, M. Reichert, and P. Dadam. On Dealing with Structural Conflicts between Process Type andInstance Changes. In Proc. Int’l Conf. BPM’04, pp. 274–289, Potsdam, 2004

S. Rinderle, M. Reichert, and P. Dadam. Disjoint and Overlapping Process Changes: Challenges,Solutions, Applications. In Proc. Int’l Conf. CoopIS’04, pp. 101–120, Ayia Napa, Cyprus, October 2004

S. Rinderle, M. Reichert, and P. Dadam. Flexible Support of Team Processes by Adaptive WorkflowSystems. Distributed and Parallel Databases, 16(1):91–116 (2004)

S. Rinderle, M. Reichert, and P. Dadam. Correctness Criteria for Dynamic Changes in Workflow Systems– A Survey. Data and Knowledge Engineering, Special Issue on Advances in Business ProcessManagement 50(1):9–34 (2004)

References(1)

v. d. Aalst, W.: Exterminating the dynamic bug: A concrete approach to supportworkflow change. Information Systems Frontiers 3:279-317 (2001)

[Aals01]

Agostini, A.; De Michelis, G.: Improving Flexibility of Workflow ManagementSystems. In Proc. BPM 2000, LNCS 1806, Springer (2000), pp. 218-234

[AgMi00]

Casati, F.; Ceri, S.; Pernici, B.; Pozzi, G.: Workflow evolution. Data andKnowledge Engineering 24: 211-238 (1998)

[CCPP98]

Edmond, D.; ter Hofstede, A.: A Reflective Infrastructure for WorkflowAdaptability. Data and Knowledge Engineering 34: 271-304 (2000)

[EdHo00]

Ellis, C.; Keddara, K., Rozenberg, G.: Dynamic change within workflow systems.In: Proc. Int‘l Conf. on Org. Systems (COOCS ‚95), Milpitas pp. 10-21 (1995)

[EKR95]

Fent, A.; Reiter, H.; Freitag, B.: Design for change,: Evolving workflowspecifications in ULTRAflow. In: Proc. CAISE ‘02, pp. 516-534 (2002)

[FRF02]

v. d. Aalst, W.; v. Hee, K.: Workflow Management. MIT Press (2002)[AaHe02]

v. d. Aalst, W.; Basten, T.: Inheritance of workflows: An approach to tacklingproblems related to change. Theoretical Computer Science 270: 125-203 (2002)

[AaBa02]

References(2)

Reichert, M.; Rinderle, S.; Dadam, P.: On the Common Support of Workflow Typeand Instance Changes Under Correctness Constraints. In: Int‘l Conf. CoopIS’03,Catania, Sicily, November (2003)

[RRD03a]

Rinderle, S.; Reichert, M.; Dadam, P.: Flexible Support Of Team Processes ByAdaptive Workflow Systems.Distributed and Parallel Databases 16(1): 91-116 (2004).

[RRD04]

Rinderle, S.; Reichert, M.; Dadam, P.: Evaluation of Correctness Criteria ForDynamic Workflow Changes. In: Int‘l Conf. on BPM‘03, Eindhoven, TheNetherlands, June (2003)

[RRD03]

Reichert, M.; Dadam, P.: ADEPTflex – supporting dynamic changes of workflowswithout losing control. Journal of Intelligent Inf. Systems 10 (1998) 93-129

[ReDa98]

Kradolfer, M.; Geppert, A.: Dynamic workflow schema evolution based onworkflow type versioning and workflow migration. In: Proc. CoopIS, Edinburgh pp.104-114, (1999)

[KrGe99]

Kochut, K.; Arnold, J.; Sheth, A.; Miller, J.; Kraemer, E.; Arpinar, N.; Cardoso, J.:IntelliGEN: A distributed workflow system for discovering protein-proteininteractions. Distributed and Parallel Databases 13:43-72 (2003)

[KAS03]

Joeris, G.; Herzog, O.: Managing evolving workflow specifications. In: Proc. Int‘lConf. on Coop. Inf. Systems (CoopIS ‘98), New York City, pp. 310-321 (1998)

[JoHe98]

References(3)

Rinderle, S.; Reichert, M.; Dadam, P.: Correctness Criteria for Dynamic WorkflowChanges – A Survey. Data & Knowledge Engineering, Special Issue onAdvances in Business Process Management 50(1): 9-34 (2004).

[RRD04a]

Weske, M.: Formal foundation and conceptual design of dynamic adaptations ina workflow management system. In: Proc. 34th Hawaii Intl‘l Conf. on SystemSciences (HICSS-34) (2001)

[Wesk01]

Rinderle, S.; Reichert, M.; Dadam, P.: On Dealing With Structural ConflictsBetween Process Type and Instance Changes. 2nd Int‘l Conf. On BusinessProcess Management, Potsdam, June 2004.Sadiq, A.; Marjanovic, O.; Orlowska, M.: Managing change and time in dynamicworkflow processes. Int‘l Journal Coop IS 9 (2000)

[RRD04b]

[SMO00]