An Approach for the Automated Composition of BPEL Processes

1
An Approach for the Automated Composition of
BPEL Processes

• Annapaola Marconi
• ITCirst / University of Trento
• marconi_at_irst.itc.it

Joint work with P. Bertoli, M. Pistore and P.
Traverso
wscomps05 Compiegne -19 September 2005
2
Outline

• Web Service composition
• Our automated composition framework
• Experimental evaluation
• Conclusions

3
Web Service Composition

• Combine existing services, available on the Web,
  to define higher level functionalities.

reqs
...
...
...
...
End user
NewExecutableCompositeProcess
WebServiceProtocols
4
Web Service Composition

• Combine existing services, available on the Web,
  to define higher level functionalities.

VIRTUAL TRAVEL AGENCY
FLIGHT
...
CUSTOMER
...
HOTEL
...
End user
WebServiceProtocols
NewExecutableCompositeProcess
5
Automated Web Service Composition

• The main idea
• Exploit and extend advanced planning techniques
  to support the automated composition of composite
  web services

planning goal
Plan
Planning
planning domain
6
Automated Web Service Composition

• The planning challenges
• Non-atomic services (complex control flow),
  nondeterministic domain

FLIGHT
register (UserID)
ok
invalid
request (Time, Loc)
not-avail
7
Automated Web Service Composition

• The planning challenges
• Non-atomic services (complex control flow),
  nondeterministic domain
• Managing exchanged data (complex data flow)

8
Automated Web Service Composition

• The planning challenges
• Non-atomic services (complex control flow),
  nondeterministic domain
• Managing exchanged data (complex data flow)

9
Automated Web Service Composition

• The planning challenges
• Non-atomic services (complex control flow),
  nondeterministic domain
• Managing exchanged data (complex data flow)

10
Automated Web Service Composition

• The planning challenges
• Non-atomic services (complex control flow),
  nondeterministic domain
• Managing exchanged data (complex data flow)

11
Automated Web Service Composition

• The planning challenges
• Non-atomic services (complex control flow),
  nondeterministic domain
• Data modeling and reasoning, complex data flow
• Complex composition requirements temporal
  preferences

try to sell holiday packages upon failure, do
never a single commit
12
Automated Web Service Composition

• The planning challenges
• Non-atomic services (complex control flow),
  nondeterministic domain
• Data modeling and reasoning, complex data flow
• Complex composition requirements temporal
  preferences

try to sell holiday packages upon failure, do
never a single commit
13
Automated Web Service Composition

• The planning challenges
• Non-atomic services (complex control flow),
  nondeterministic domain
• Data modeling and reasoning, complex data flow
• Complex composition requirements temporal
  preferences

try to sell holiday packages upon failure, do
never a single commit
14
Outline

• Web Service composition
• Our automated composition framework
• Experimental evaluation
• Conclusions

15
Our service composition framework
?
Compositionrequirements
W
Synthesizedplan
W1
ExecutableCompositeService
. . . .
. . . .
Planning domain
Wn
State TransitionSystems
ComponentWeb Services

• Component services abstract BPEL4WS semantic
  annotations
• Composite service executable BPEL4WS
• Composition requirements EaGLe requirements
  language

16
Abstract BPEL4WS Semantic Annotations

• BPEL4WS complete and detailed description of the
  WS protocol
• Automated composition task needs a semantic
  description of the interactions
• Semantic annotations
• define the outcome of the interaction

17
Abstract BPEL4WS Semantic Annotations

• BPEL4WS complete and detailed description of the
  WS protocol
• Automated composition task needs a semantic
  description of the interactions
• Semantic annotations
• define the outcome of the interaction
• add semantic relations among the exchanged data

18
Abstract BPEL4WS Semantic Annotations

• BPEL4WS complete and detailed description of the
  WS protocol
• Automated composition task needs a semantic
  description of the interactions
• Semantic annotations
• define the outcome of the interaction
• add semantic relations among the exchanged data

19
EaGLe requirements language

• EaGle (Dal Lago, Pistore, Traverso 2002)
• a language for temporally extended planning goals
  in non-deterministic domains
• many similarities with CTL
• recovery conditions, second-preference goals

20
EaGLe requirements language
EaGLe formalization
TryReach C.booked ? H.booked ? F.booked ?
C.cost F.costOf(C.time, C.loc)
H.costOf(F.time, C.loc) Fail DoReach
C.not_booked ? H.not_booked ? F.not_booked
try to sell holiday packages upon failure, do
never a single commit
21
Our service composition framework
?
Compositionrequirements
W
Synthesizedplan
W1
ExecutableCompositeService
. . . .
. . . .
Planning domain
Wn
State TransitionSystems
ComponentWeb Services

• Component services abstract BPEL4WS processes
• Composite service executable BPEL4WS process
• Composition requirements EaGLe requirements
  language

22
Our service composition framework
?
Compositionrequirements
W
Synthesizedplan
W1
WS2STS
ExecutableCompositeService
. . . .
. . . .
Planning domain
Wn
ComponentWeb Services
State TransitionSystems

• WS2STS from web services to formal models
• support all BPEL4WS basic and structured
  activities
• !!! deal with DATA (knowledge level planning)

23
Knowledge Level Planning

• Reasoning on the data values exchanged by the web
  services participating to the composition
• Remark neither actual data values nor actual
  definition of functions are important!
• Key challenge define a suitable knowledge base
  s.t.
• KL models can be automatically extracted from
  BPEL processes
• automated composition can be done for a relevant
  set of realistic problems
• composite BPEL process can be automatically
  extracted from the plan
• efficient composition
•  

24
Knowledge Level Planning

• Knowledge Base Propositions
• KV(x) we know the value of x
• K(x y) we know that x and y have the same
  value
• K(x f (y1, .., yn)) we know that x has the
  value of f applied to y1, .., yn
• where x, y, y1, .., yn are BPEL
  variables and f is a user defined function. 

• Knowledge Base Actions
• del (KB, p1, .., pn) is the knowledge base KB \
  p1, .., pn
• add (KB, p1, .., pn) is the knowledge base KB U
  p1, .., pn
• close (KB) is the knowledge base containing all
  propositions that can be deduced from the
  propositions in KB

25
Knowledge Level Planning

• KB evolution a simple example

• Suppose that 
• r, v, z Request and o, w Offer are BPEL
  variables
• costOf(Request, Offer) is a BPEL function (of
  the Hotel WS)
• KB0 KV(v) , K(r z) is the current
  knowledge base
• Are executed the following actions
• request (r ? v)
• o costOf (r)
• offer(o ? w)

26
Knowledge Level Planning

• KB evolution a simple example

KB0 KV(v) , K(r z)

• request(r ? v) on KB0
• preconditions
• KV(v)
• effects
• KB0 del (KB0 , ..all props on r) KV(v)
• KB0 add (KB0 , K(r v)) KV(v), K(r
  v)
• KB1 close (KB0 ) KV(v), K(r v),
  KV(r)

27
Knowledge Level Planning

• KB evolution a simple example

KB1 KV(v), K(r v), KV(r)

• o costOf(r) on KB1
• preconditions
• none
• effects
• KB1 del (KB1 , ..all props on o) KV(v),
  K(rv), KV(r)
• KB1 add (KB1 , K(o costOf(r)))
  KV(v), K(rv), KV(r), K(ocostOf(r))
• KB2 close (KB1 ) KV(v), K(rv), KV(r),
  K(ocostOf(r)), K(ocostOf(v))

28
Knowledge Level Planning
KB2 KV(v), K(r v), KV(r),
K(ocostOf(r)), K(ocostOf(v))

• KB evolution a simple example

• offer(o ? w) on KB2
• preconditions
• none
• effects
• KB2 del (KB2 , ..all props on w..)
  KV(v), K(rv), KV(r), K(ocostOf(r)),
  K(ocostOf(v))
• KB2 add (KB2 , K(w o), KV(w)) KV(v),
  K(rv), KV(r), K(ocostOf(r)),
  K(ocostOf(v)), K(wo), KV(w)
• KB3 close (KB3 ) KV(v), K(rv), KV(r),
  K(ocostOf(r)), K(ocostOf(v)), K(wo),
  KV(w), KV(o), K(wcostOf(r), K(wcostOf(v)

29
Our service composition framework
?
Compositionrequirements
W
Synthesizedplan
W1
WS2STS
ExecutableCompositeService
. . . .
. . . .
Planning domain
Wn
ComponentWeb Services
State TransitionSystems

• WS2STS from web services to formal models
• support all BPEL4WS basic and structured
  activities
• !!! deal with DATA (knowledge level planning)

30
Our service composition framework
?
Compositionrequirements
W
Synthesizedplan
W1
ExecutableCompositeService
. . . .
. . . .
Planning domain
Wn
ComponentWeb Services
State TransitionSystems

• Obtaining the planning domain
• S is a STS representing all the possible
  behaviors of the component services

31
Our service composition framework
?
Compositionrequirements
AUTOMATED COMPOSITION
W
Synthesizedplan
W1
ExecutableCompositeService
. . . .
. . . .
Planning domain
Wn
ComponentWeb Services
State TransitionSystems

• Synthesizing the plan
• Planning via Symbolic Model Checking (MBP Planner
  - NuSMV)

32
Our service composition framework
?
Compositionrequirements
W
STS2WS
Synthesizedplan
W1
ExecutableCompositeService
. . . .
. . . .
Planning domain
Wn
ComponentWeb Services
State TransitionSystems

• STS2WS Obtaining the executable composite
  process
• ready for deployment and run
• particular care to code quality

33
Outline

• Web Service composition
• Our automated composition framework
• Experimental evaluation
• Conclusions

34
Experiment set 1 Atomic WS Composition
35
Experiment set 2 More complex WS Composition
36
Experimenting with realistic applications

• On the HF problem
• Time
• automated composition 10 secs
• manual coding 1 hour
• Quality
• automated composition 22K bytes
• manual coding 11K bytes

37
Outline

• Web Service composition
• Our automated composition framework
• Experimental evaluation
• Conclusions

38
Conclusions

• The proposed composition framework has been
  implemented within the ASTRO toolset
• http//astroproject.org

39
Conclusions

• The proposed composition framework has been
  implemented within the ASTRO toolset
• http//astroproject.org

• ASTRO toolset supports
• ..of composite business processes

FORMAL VERIFICATION
RUN-TIME MONITORING
AUTOMATED COMPOSITION
40
Some publications
http//astroproject.org

• Planning Framework for Web Service Composition
• M. Pistore, P. Traverso, P. BertoliAutomated
  Composition of Web Services by Planning in
  Asynchronous DomainsInternational Conference on
  Automated and Planning Sheduling (ICAPS) 2005
• Managing exchanged data
• M. Pistore, A. Marconi, P. Bertoli, P.
  TraversoAutomated Composition of Web Services by
  Planning at the Knowledge LevelInternational
  Joint Conference on Artificial Intelligence
  (IJCAI) 2005
• Semantic Web Service Composition
• P. Traverso, M. PistoreAutomatic Composition of
  Semantic Web Services into Executable
  ProcessesInternational Semantic Web Conference
  (ISWC) 2004
• Composition Verification
• R. Kazhamiakin, M. PistoreA Parametric
  Communication Model for the Verification of
  BPEL4WS Compositions2nd International Workshop
  on Web Services and Formal Methods (WS-FM) 2005

41
An Approach for theAutomated Composition of BPEL
Processes
42

• COMPOSITE WS VERIFICATION
• R. Kazhamiakin, M. PistoreA Parametric
  Communication Model for the Verification of
  BPEL4WS Compositions2nd International Workshop
  on Web Services and Formal Methods (WS-FM) 2005
• R. Kazhamiakin, M. Pistore, M. RoveriA framework
  for integrating Business Processes and Business
  Requirements9th International IEEE Enterprise
  Distributed Object Computing Conference (EDOC)
  2004
• M. Pistore, M. Roveri, P. BusettaRequirements-Dri
  ven Verification of Web Services1st
  International Workshop on Web Services and Formal
  Methods (WS-FM) 2004
• R. Kazhamiakin, M. Pistore, M. RoveriFormal
  Verification of requirements using SPIN A Case
  Study on Web Services3rd IEEE International
  Conference on Software Engineering and Formal
  Methods (SEFM) 2004

43

• SERVICE SYNTHESIS AND COMPOSITION
• M. Pistore, A. Marconi, P. Bertoli, P.
  TraversoAutomated Composition of Web Services by
  Planning at the Knowledge LevelInternational
  Joint Conference on Artificial Intelligence
  (IJCAI) 2005
• M. Pistore, P. Traverso, P. Bertoli, A.
  MarconiAutomated Synthesis of Composite BPEL4WS
  Web Services3rd IEEE International Conference on
  Web Services 2005
• M. Pistore, P. Traverso, P. Bertoli, A.
  MarconiAutomated Synthesis of Executable Web
  Serivce Compositions from BPEL4WS
  ProcessesPoster presentation at the
  International World Wide Web Conference (WWW)
  2005
• M. Pistore, P. Traverso, P. BertoliAutomated
  Composition of Web Service by Planning in
  Asynchronous DomainsInternational Conference on
  Automated and Planning Sheduling (ICAPS) 2005
• M. Pistore, F. Barbon, P. Bertoli, D. Shaparau,
  P. TraversoPlanning and Monitoring Web Service
  CompositionArtificial Intelligence Methodology,
  Systems, Application (AIMSA) 2004Also ICAPS'04
  Workshop on Planning and Scheduling for Web and
  Grid Services

44

• SEMANTIC WEB SERVICES
• M. Pistore, P. Roberti, P. TraversoProcess-Level
  Composition of Executable Web Services
  "On-the-fly" Versus "Once-for-all"
  CompositionEuropean Semantic Web Conference
  (ESWC) 2005
• P. Traverso, M. PistoreAutomatic Composition of
  Semantic Web Services into Executable
  ProcessesInternational Semantic Web Conference
  (ISWC) 2004