Presentation is loading. Please wait.

Presentation is loading. Please wait.

WAIM’01- Slide 1 Web-based Workflow Evolution in ADOME-WFMS Dickson K.W. CHIU Dept. of Computer Science, Hong Kong University of Science.

Similar presentations


Presentation on theme: "WAIM’01- Slide 1 Web-based Workflow Evolution in ADOME-WFMS Dickson K.W. CHIU Dept. of Computer Science, Hong Kong University of Science."— Presentation transcript:

1 WAIM’01- Slide 1 Web-based Workflow Evolution in ADOME-WFMS Dickson K.W. CHIU (kwchiu@ieee.org) Dept. of Computer Science, Hong Kong University of Science & Technology Qing LI (csqli@cityu.edu.hk) Dept. of Computer Science, City University of Hong Kong Kamalakar KARLAPALEM (kamal@iiit.net) Indian Institute of Information Technology

2 WAIM’01- Slide 2 Workflow Evolution Features in ADOME WFMS Meta-modeling approach Simplicity and expressiveness Modeling and implementation tied together in a straightforward manner Web Interface for workflow evolution Support ECA-rule based handover policies

3 WAIM’01- Slide 3 Three-Level Meta-Modeling for ADOME-WFMS

4 WAIM’01- Slide 4 Meta-level Specification of Activities and Tasks class Task isa Activity attribute_initilization: Activity_Graph = NIL; attributes: Task_Need: set of Token; Allow_Partial_Match: bool; methods Match_Cost, Partial_Match, Match_Cost_Ext... end class Arc isa WFMS_class Source, Dest : Activities; Trans_Condition: Boolean; end; class Graph_Node isa WFMS_class attributes: Subactivity: Activity; Predecessor_Join: (AND_join, OR_join, NIL); Successor_Split: (AND_split, OR_split, NIL); end; class Activity isa WFMS_class /* all activities are sub-classes of this meta-activity class because each different activity class can have multiple instance */ class_attributes: Input_Parameters, Output_Parameters, IO_Parameters: set of Parameter; events: execute, finish, abort... exceptions: no_PSA, PSA_reject, cannot_proceed rules: Manditory_handlers: set of rules; Handlers: set of rules; attributes: Activity_Graph: (Activity_Node set of Graph_Node; Transisition set of Arc;) Reexecution_Pattern: (optional, repeatable, replacable…); /* parameters for execution of instance */ Priority: integer; PSA_chosen: set of PSA; methods Decomposition, PSA_for_Activity, Execution,.... end

5 WAIM’01- Slide 5 Workflow Evolution High-level / User-oriented workflow evolution actions –Exception avoidance –Comprise workflow evolution primitives Workflow Evolution Primitives Workflow Evolution Policies – how executing workflow instances adopt newly evolved workflow schema

6 WAIM’01- Slide 6 Workflow Evolution in an OO WFMS Changing a workflow instance – changing the content of a workflow object Changing a workflow definition – changing a workflow class schema (supported by ADOME active OODBMS facilities)

7 WAIM’01- Slide 7 Workflow Evolution Primitives Declaration primitives - modify declaration of variables (including their default values) –RemoveVar (a: Activity, v: Variable_Name) –Addvar (a: Activity, v: Variable_Name, t: Type, d: DefaultValue) Flow primitives - modify the flow structure of the workflow schema –AppendNode (n, a: Activity) –SplitNode (n, a: Activity; b: Boolean) –JoinNode (a: Activity; w: set of Activity) –ModifyCondition(s,t: Activity; b: Boolean) Common Operators derived from other primitives: –InsertBefore (n, a: Activity) –DeleteNode (n : Activity)

8 WAIM’01- Slide 8 Some Algorithms for Workflow Evolution Primitives Direct and simple implementation based on meta-workflow definition PROCEDURE AppendNode (n, a: Activity, G: Activity_Graph) IF (a=END) OR (n IN G.Subactivity) THEN RETURN ERROR; G.Arc = G.Arc U (a, n, true) G.Graph_node = G.Graph_node U (n, NIL, a.Successor_Split) FOREACH j in G.Arc where j.Source=a j = (n, j.Destination, j.Transition_Condition) END PROCEDURE SplitNode (n, a: Activity, b: Boolean, G: Activity_Graph) IF (n=a) OR (n IN a.Successor) OR (a.Successor_Split=AND_Split) OR (n.Predecessor_Join=AND_join ) THEN RETURN ERROR G.arc = G.arc U (a, n, b) a.Successor_Split = OR_split; n.Predessor_Join = OR_join END PROCEDURE ModifyCondtion(s, t: Activity, b: Boolean; G: Activity_Graph) IF (s, t, *) not IN G.arc THEN RETURN ERROR UPDATE j.Transition_Condition = b WHERE j IN G.arc AND j.source=s AND j.destination=t END

9 WAIM’01- Slide 9 Sample Specification for High-level / User-oriented Workflow Evolution Actions in ADOME-WFMS

10 WAIM’01- Slide 10 High Level / Application Oriented / User Defined Workflow Evolution Actions

11 WAIM’01- Slide 11 Based on workflow evolution primitives

12 WAIM’01- Slide 12 Workflow Evolution Policies Temporary change: only change the current instance (e.g., as a trial). Delayed change: change the current and future instances Abrupt change: abort the current and other executing instances of the sub-activity; re-execute with the new definition or alternate paths as determined by the exception handling mechanism for the parent activity (e.g., because of a critical problem) Immediate change: all the instances suddenly use the new definition (e.g. adding new alternative branches / exception handler) Progressive change: allow different instance to take different decisions on whether to hand over to the new schema. (This needs most attention.)

13 WAIM’01- Slide 13 Progressive Workflow Evolution based on Event-Condition-Action (ECA) rules Besides general expressions, some interesting primitives / expressions used in specifying rules for progressive workflow evolution are: Event – position (current task) of a running instances Condition – past execution history Action –rollback to t – undo some work back to the execution position t before adopting new schema –change over to t – migration to the new schema by jumping to position t –go ahead – continue to use old schema

14 WAIM’01- Slide 14 More flexible ECA-rule based handover policy specification Higher level handover policy

15 WAIM’01- Slide 15 Selected Reference (=> http://www.cs.ust.hk/~kwchiu) D.K.W. Chiu, Q. Li and K. Karlapalem,. A Meta Modeling Approach for Workflow Management System Supporting Exception Handling. Information Systems, Pergamon Press, Elservier Science, 24(2):159-184, 1999. D.K.W. Chiu, Q. Li and K. Karlapalem. Web Interface-Driven Cooperative Exception Handling in ADOME Workflow Management System, Information Systems, Pergamon Press, Elservier Science 26(2), 2001 D.K.W. Chiu, Q. Li and K. Karlapalem. ADOME-WFMS: Towards Cooperative Handling of Workflow Exceptions, in Advances in Exception Handling Techniques, A. Romanovsky, C. Dony, J. L. Knudsen, A. Tripathi (Eds), Springer-Verlag, LNCS-2022, 2001 D.K.W. Chiu, Q. Li and K. Karlapalem. Views for Inter-Organization Workflow in an E-Commerce Environment, DS-9, Hong Kong, April 2001 D.K.W. Chiu, Exception Handling in an Object-Oriented Workflow Management Systems, PhD Thesis, Dept of Comp. Sci., HKUST, Aug 2000. D.K.W. Chiu. Workflow Management Systems as Meta-Agents in a Multi- Agent Information System, 3rd International Workshop on Agent Oriented Information Systems (AOIS-2001), June 2001


Download ppt "WAIM’01- Slide 1 Web-based Workflow Evolution in ADOME-WFMS Dickson K.W. CHIU Dept. of Computer Science, Hong Kong University of Science."

Similar presentations


Ads by Google