IG-JADE-PKSlib: An Agent-Based Framework for Advanced Web Service Composition and Provisioning

541 views
464 views

Published on

We describe an agent-based infrastructure and toolkit to develop inter-operable, intelligent, multiagent systems for Web service composition (WSC) and provisioning. Our toolkit is realized through an interface library (IG-JADE-PKSlib) that combines state of the art agent-based and planning technologies (i.e., the IndiGolog model-based agent programming language, the JADE agent platform, and the PKS planning system). We show that each of these tools has its strengths and weaknesses, but combined together, they provide a very powerful toolkit. We argue that this infrastructure is particularly well suited for developing next generation Web services (WS) applications.

Published in: Technology, Education
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
541
On SlideShare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
8
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

IG-JADE-PKSlib: An Agent-Based Framework for Advanced Web Service Composition and Provisioning

  1. 1. IG-JADE-PKSlib An Agent­Based Framework for  Advanced Web Service Composition  and Provisioning  Erick Martínez & Yves Lespérance Department of Computer Science York University Toronto, Canada1 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  2. 2. Motivation ● Next generation Web Services – Semantic Web – Reasoning / Planning – Automation ● Agent­oriented toolkit for advanced MAS /  WSC and provisioning ● Previous results (knowledge­based  planning)2 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  3. 3. Air Travel MAS Example (WSC & Provisioning)3 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  4. 4. IG-JADE-PKSlib Components ● IndiGolog: high­level model­based  programming language based on the situation  calculus and good for modeling dynamic  domains. ● JADE: open source, Java­based, FIPA­ compliant software framework for developing  MAS ● PKS: efficient knowledge­based planning  system, supports incomplete information and  sensing actions4 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  5. 5. IndiGolog (IG) (De Giacomo & Levesque, 1999) ● Supports: – Reasoning about action (e.g., diagnosis)  – Exogenous events handling, – Sensing,  – Planning / Re­planning,  – Plan execution, – Execution monitoring ● Limitations: – mainly intended for designing individual  autonomous agents5 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  6. 6. IG Agent Specification Has two parts: ● Declarative specification of the domain and its dynamics  in the situation calculus: – Axioms describing the initial situation – Action precondition axioms – Successor state axioms – Sensed  fluent axioms – Unique names axioms for the primitive actions – Some foundational, domain independent axioms  ● Procedural description of the agent behaviour in the IG  process language – Rich programming language, with support for non­determinism,  concurrency, interrupts, search, etc.6 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  7. 7. IG Travel Orchestrator Agent Example proc orchControl     π id, pid, xid .        ?(agent_ID = id),       ?(plan_ID = pid),       ?(exec_ID = xid),    /* Three Concurrent Processes */    handleRequest(id) ||    handlePlanning(pid) ||    handleExecution(xid) endProc7 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  8. 8. Java Agent DEvelopment Framework (JADE) (TILAB/AOT Labs) ● Supports: – FIPA­compliant agent architecture ● Agent Management System ● Directory Facilitator ● Agent Communication Channel – Full implementation of FIPA communication model ● Agent interaction protocols (AIPs),  ● Agent Communication Language (ACL),  ● SL content language,  ● Some types of ontologies, and ● Transport protocols ● Limitations: – Lack of reasoning mechanisms for agents ● Interface to JESS, but JESS has limitations with respect to  reasoning about dynamic domains (e.g., WS)8 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  9. 9. IG-JADE Interface IndiGolog Exogenous Primitive Actions (IG) TCP/IP Socket Actions (IG) JADE9 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  10. 10. IG-JADE Interface (IG Arch.) ● Original IG  Limitation:  intended  for designing individual  autonomous agents ● Extended IG Features: – FIPA ACL (message)  handling – FIPA SL (content)  handling  – Support for FIPA Agent  Interaction Protocols  (AIPs) ● Integration to MAS ● Asynchronous  communication – Proactive + reactive IG  agents10 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  11. 11. IG Travel Orchestrator Agent Example (Revisited) proc handleRequest(id)    /* FIPA AIP (Request­participant) */    initRequestParticipantRole(id) endProc proc orchControl   proc handlePlanning(pid)      π id, pid, xid .     <cid: state_planning(pid, cid) = ready         ?(agent_ID = id),        /* FIPA AIP (Request­initiator) */         ?(plan_ID = pid),        → initRequestInitiatorRole(id)>       ?(exec_ID = xid),    〉〉     handleRequest(id) ||    handlePlanning(pid) ||        <true →  noOp>    handleExecution(xid) endProc endProc11 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  12. 12. IG-JADE Interface (JADE Arch.) ● Original JADE  Limitation:  lack of  reasoning mechanisms  for agents (dynamical  domains) ● Extended JADE  Features: – Socket interface to IG ● Messenger is a full­ fledged JADE agent – ACL handling – Translator/mailbox ● Extended capabilities  to do reasoning on  dynamical domains12 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  13. 13. Air Travel MAS Example (Revisited)13 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  14. 14. Automated Web Service Composition as Planning ● WSC Problem: given a set of Web services and some  user defined task or goal to be achieved, automatically  find a composition of the available services to accom­ plish the task. ● WSC as a Planning Problem:   – Predefined available services as the building blocks of a plan – Many WS actions involve sensing – Large search space, incomplete information in the initial  state14 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  15. 15. Planning with Knowledge and Sensing (PKS) ● Supports: – Planning under conditions of incomplete knowledge – Sensing actions, – Parameterized conditional plans, – Planning at the knowledge level (without considering  all the different ways the physical world can be  configured), – Efficient, sound, forward chaining inference algorithm ● Limitations: – Only limited forms of incomplete knowledge can be  represented  – Inference algorithm is efficient but incomplete15 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  16. 16. WSC in PKS (Martínez & Lespérance, 2004) ● PKS primitive actions correspond to WS – Knowledge­producing actions  ↔    information  gathering WS – Physical actions  ↔   world­altering WS ● New WS becomes available  →    add new  primitive action to domain specification16 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  17. 17. PKS Spec. Example (Air Travel Domain) Action Precondition Effects  findRFlight(x)  K(airCo(x))  add(Kw, flightExists(x))  K(indFindRFlight(x))  add(Kf,¬indFindRFlight(x))  K(desFindRFlight(x))  bookFlight(x)  K(airCo(x))  add(Kf, bookedFlight(x))  K(availFlight(x))  del(Kf, availFlight(x))  K(indBookFlight(x))  add(Kf,¬indBookFlight(x))  K(desBookFlight(x)) Domain specific update rules (DSUR)  K(airCo(x))  ¬Kv(flightNum(x))  K(flightExists(x))   add(Kv,  flightNum(x)) 1 explicit parameter: company17 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  18. 18. PKS-JADE Interface ●  Why PKS­JADE?  – Both object­oriented: JADE (Java), PKS (C++)  ● PKS­WA is a full­fledged JADE agent  – FIPA AIPs – FIPA ACL – FIPA SL – Ontology support (e.g., pks­ontology) ● Extended knowledge­based planning capabilities18 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  19. 19. Air Travel MAS Example (Final / IG-JADE-PKSlib)19 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  20. 20. Main Contributions ● Extended IndiGolog, added support for: – FIPA Agent Interaction Protocols (AIPs) – FIPA ACL (message) handling – FIPA SL (content) handling  – Integration to MAS ● Asynchronous communication, proactive + reactive IG agent  behaviours ● IG­JADE­PKSlib: agent­based infrastructure and toolkit, for  developing inter­operable, intelligent MASs (e.g. advanced WS  applications) ● Our toolkit provides comprehensive support for multiple fea­ tures, such as: reasoning for dynamical domains, knowledge­ based planning (incomplete knowledge, sensing), execution  monitoring, AIPs, ontologies, etc.20 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  21. 21. Open Problems and Future Work ● Translation of OWL, DAML­S, etc. into PKS/IG specifications ● Representation of atomic services ● Customizing domain theory based on problem – Automatic generation (desc. goal + user pref.) ● IG­PKS interface (i.e., IG built­in KB planning) – IG passes planning calls to PKS (efficient) – Prolog­based reimplementation of PKS (more flexible, but less  efficient)  ● Experiments + case studies to validate performance and     scalability of integrated framework [Martínez & Lespérance, 2004] ● Plan execution and contingency recovery21 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004
  22. 22. Thank You!22 Martínez & Lespérance IG-JADE-PKSlib WSABE @ AAMAS-2004

×