Web Service Composition as Planning

1

Introduction to Artificial Intelligence
MSc 2005 Semester 2 8th Tutorial

• Web Service Composition as Planning?

2
The problem

• Is composition of Web Services possible by
  describing services as planning
  operators/domains?
• Are current planners suitable for this kind of
  composition?
• 1000s of services, probably doing almost the same
  (e.g. there are loads of Hotel booking services)?

3
a pessimistic view from the lecture slides
Jim Hendler's opinion 01/02/2004
public-sws-ig_at_w3.org

• I would point out that the mapping of web
  services to compositions has largely been done in
  the past, even in the best work in
• this area, with some simplifications that
  generally "twist" web services into a planning
  framework -- there's huge parts of the
• web service world that need to be explored before
  we can really say AI planning has shown a success
  in web services other
• than as an evocative idea -- the reason is that
  a real web services engine will need to deal with
  (at least)
• scaling issues way beyond anything we've seen in
  planning to date (there may be thousands of
  services each with multiple ports, optional
  arguments, etc.)
• the issues Dana mentioned (side effects, change
  in the world, etc.) that make Strips-operators
  planning an approximation at best (the assumption
  that change occurs only through the operators
  under the control of this planner is clearly
  wrong)
• issues of interaction with users - web service
  planning better be more mixed-initiative
• issues of preferences v. constraints
• issues of interaction between planning agents out
  there (you buy the book I'm in the process of
  planning to buy)
• knowledge engineering issues (when planners take
  ebXML and WSDL as inputs, instead of requiring
  specialized planning-like languages like
• OWL-S, then we'll see a lot more excitement on
  the outside - OWL-S is an interesting starting
  place, but we fool ourselves if we think it
  really is going to be widely used for process
  specification in its current form)
• In fact, I'll wager that it will be absolutely
  trivial to prove that web services planning, even
  w/simplifications, is inherently
• undecidable, so we'll need to explore a lot of
  the issues from the old "dynamic planning" world
  as well. All this, by the way, I
• see as good news - it means this is a fertile and
  exciting research area for those of us in
  planning, with good heuristic solutions
• being transitionable. That said, in the past I
  have tried to get AI planning people to think
  outside the box and failed miserably,
• and I'll be surprised if the web services
  planning stuff doesn't become an "applied" area
  being ignored by the bulk of the
• research community (who, if you'll apologize my
  saying so, still have their heads up their
  you-know-whats worrying about
• scaling simple problems in all the wrong ways)

4
Classical Planner are not necessary suitable

• BUT Can serve as an approximation at a high
  level of abstraction in semi-automatic
  approaches!
• i.e. planning as a "heuristics" for WSC!
• You should try out this week how far this is
  possible and under which restrictions!

5
Assignment this week

• This weeks assignment is a challenge for your
  creativity!
• You should
• describe a service composition problem as a
  planning problem in PDDL
• Generate a big set of operators, emulating lots
  of possible similar services and observe
  scalability
• Find a suitable planner for your problem, run it.
• If not possible to describe/solve your encoding,
  find usable abstractions, simplifications.
• ? See last slide for the concrete task.

6
Web and Services

• Text/images information providing/world
  altering services
• Manual Web services interoperation
• Agents interoperation Semantic services
• Service markup machine understandable

7
Web and Services

• Text/images information providing/world
  altering services
• Manual Web services interoperation
• Agents interoperation Semantic services
• Service markup machine understandable
• We try to use PDDL this week to semantically
  describe Web services!

8
Semantic Service Description

• Problem/Vision
• automatic discovery,
• execution and
• composition of Web services
• Solution
• Semantic markup of Web services
• in our case we will PDDL
• More general approach Service ontologies, OWL-S,
  WSMO

9
Service Composition andInteroperation

• The automatic selection, composition, and
  interoperation of appropriate Web services to
  perform some task, given a high-level description
  of the tasks objective.
• Make the travel arrangements for my IJCAI 2001
  conference trip.
• Can be formulated as a simple goal in PDDL

10
Industrial Standards

• UDDI, WSDL, ebXML
• E-speak (HP), .NET (MS)
• For us here most important
• WSDL exposes operations describing how to
  interact with a service, but not really what it
  does.

11
Example XMethods

• www.xmethods.com a repository of real Web
  Services available online
• E.g. Global Weather service
• http//www.webservicex.com/globalweather.asmx?WSD
  L

Most services currently available are as simple
as simple procedure calls.
12
Problem with XMethods

• WSDL only describes in-out of service operations,
  but not what the service does (preconditions,
  effects are missing)
• Most services, such as the weather service are
  pure information gathering services.
• You gather information as you execute the action,
  the outcome is non-deterministic, i.e. we don't
  know what
• getWeather(Innsbruck)
• will return.
• Current planners are not conceived for
  information gathering, i.e. adding knowledge or
  taking decisions based upon the outcome of
  actions.
• More interesting, in our context are services
  with real world effects, such as charging my
  credit card, etc.
• Since existing services lack these descriptions,
  let us use some fictitious but useful web
  services, e.g. Hotel booking, car rental, flight
  booking, etc.

13
Sample Problem Online Trip Booking

• Let us now try to encode a simple problem and a
  few services as a planning problem in PDDL
• I have 100 on my credit card and 150 on my bank
  account and I want to purchase a ticket way and
  back to Vienna and book a hotel in Vienna.
• Initial State and Goal
• (define (problem ticketAndHotelVienna)
• (domain servicesPlanning)
• (objects
• Innsbruck Vienna city
• 2000 2001 2002 2004 2005 - year
• (init ( (creditCardBalance) 100)
• ( (accountbalance) 150)
• ( (total-money-spent) 0)
• )
• (goal(and (haveTicket Innsbruck Vienna)
• (haveTicket Vienna Innsbruck)
• (haveHotelReservation Vienna))
• (metric minimize (total-money-spent))
• )

14
Online Trip Booking - Domain
fluents allow function definitions and use
of assignments and arithmetic expr. In effects

• (define (domain servicesPlanning)
• (requirements typing strips fluents)
• (types city year)
• (predicates (haveTicket ?x ?y - city)
• (haveReservation ?x - city))
• (hasVorteilsCard ?y - year)
• (functions (total-money-spent)
• (accountBalance)
• (creditCardBalance))
• (action TheRottenHotelBookingService
• parameters (?x - city)
• preCondition (gt (creditCardBalance) 50)
• effect (and (haveReservation ?x)
• (decrease (creditCardBalance)
  50)
• (increase (total-money-spent
  50))))
• (action TheFancyHotelBookingService
• parameters (?x - city)

15

• (action onlineBankingService
• parameters (?transferAmount)
• preCondition (gt (accountCardBalance)
  ?transferAmount)
• effect (and (decrease (accountBalance)
  ?transferAmount)
• (increase (creditCardBalance)
  ?transferAmount)))
• (action oebbTicketService
• parameters (?from ?to)
• effect (and (hasTicket ?from ?to)
• (decrease (creditCardBalance)
  70)
• (increase (total-money-spent
  70))))
• (action oebbVorteilsCardTicketService
• parameters (?from ?to)
• preCondition (and (hasVorteilsCard 2005)
  (gt (accountCardBalance) 20 )
• effect (and (hasTicket ?from ?to)
• (decrease (creditCardBalance)
  20)
• (increase (total-money-spent
  20))))

16
Your exercises

• Improve the very simple encoding in the slides,
  e.g. by
• parametrizing by date, number of days to stay in
  the hotel
• The vorteilscard has to be valid for the date of
  travel,
• adding more services,
• adding more cities and different costs for
  tickets encoded in some fixed predicates,
• etc.
• Think of an own example in the web service
  composition area and model it in PDDL. Try to
  avoid simplifications and keep your modeling as
  realistic as possible. You can use any extension
  of PDDL which you might find useful. What are the
  obstacles?
• Find a planner which can solve your problem.
• Now try to enrich you modeling by adding more
  services (actions) which offer the same or only
  slightly different functionality, does it affect
  scaling?
• Answer the following questions
• Which abstractions, simplicifcations did you have
  to make in your modeling to tweak the composition
  problem into PDDL?
• Is it feasible that Web services advertise their
  functionality in a language like PDDL? Why? Why
  not?
• Why can't you use WSDL directly for planning and
  composition, what is missing?
• Relate your observations to assumptions A0-A7
  from the lecture slides.
• As mentioned in the beginning This exercises is
  a challenge for your creativity and to find out
  what are the borders of current languages and
  technologies.
• Due to travelling and public holidays, you have
  more time than usual for this assignment