Ontology-Driven Web Services
Composition Platform
I. Budak Arpinar
Boanerges Aleman-Meza *
Ruoyan Zhang
Angela Maduko
Large Scale Distributed Information Systems Lab
Computer Science Dept., University of Georgia
1
Web Service
• A Web Service is a software application identified by a
URI, whose interfaces and binding are capable of being
defined, described and discovered by XML artifacts and
supports direct interactions with other software
applications using XML based messages via Internetbased protocols (W3C definition)
• A self-contained,
self-described,
and self-advertised
composition unit
Web Services Stack
Service Publication/
Discovery UDDI
Service Description WSDL
XML Messaging SOAP
Transport Network HTTP
2
Web Services enabling E-Commerce
Service Registry
(UDDI)
Send Request
(XML)
Publish WS
Service Provider
Customer
Bind
Receive Form
Organization A
SOAP
Check Postcode
Organization B
3
Web is turning into a collection of Web
Services
• The number of companies that
have completed an IT project
involving Web services standards
has grown (survey released in
2003 [TechWeb])
• Surveyed firms using standards
such as XML or SOAP increased
from 11% percent in mid-2002 to
31% percent 2003 (market
researcher Forrester Research Inc)
4
Globalization of Web processes
B2B
Enterprise
Workflows
Enterprise
Distributed
Workflows
Inter-Enterprise
E-Services
Web Processes
(Composition)
Global
[Sheth, Cardoso WSTutorial]
5
Composition ideas
Past/ Manual
Now /Semi-Automatic
Web Service
Ontology
Web Service
Ontology
S
S
S
S
s
Ws
Instance
Semantic
Functionality
S S
S S
S
S
S
S s
S
S
S
Richer
relationship
S
S
S S
S
S S
s
Soon /Automatic ?
S
S
S
Ws
Instance
WS
Instance
Automatic Composition/
Possible dynamically
generated/adapted web
process
Semantics is the most important enabler to
address these challenges
Workflow/Static
Possible Dynamic
modification
Template-based Web Process
6
Architecture
Modules fit in the
METEOR–S
Framework


Semantic annotation,
discovery,
composition of WSs
Data, Operational,
Functional, and QoS
Semantics
7
Interface-Matching Composition
Human-Assisted Composition
Web Services Composition in
Peer-to-Peer networks
8
Interface-Matching Algorithm





Web services assembled in forward -chaining fashion
Semantic matching of input/outputs are computed
among WS (ontological measures are used for
matching)
A WS network/graph is generated with matching
inputs/outputs
A forward-chaining search locates an optimal
composition path
Adapted Bellman-Ford Shortest path algorithm
(involving multiple inputs and outputs)
9
Interface-Matching Composition
Human-Assisted Composition
Web Services Composition in
Peer-to-Peer networks
10
Human-Assisted Composition

Involves human for consideration of Quality
rate, cost, time, geographic region, etc

Follows a iterative Template-based
composition method

User-parameters serve as filters / ranking
11
Travel Planner Example (stage 1)
Event Information
Event Information
Flight Booking
Confirmation
Arrival Date
Transportation Booking
Service
Address (Source)
Departure
Date
Departure
time
Arrival Date
Address (Source)
Map and Driving
Direction
Location
Map Service
Check -Out Time
Check-In Time
Address(Destination)
Lodging Booking Service
Address (hotel)
Hotel Reservation
Confirmation
INPUT
OUTPUT
12
Ontology-based Web Services Selection
Selection for service classes / categories
select the appropriate subclasses of the
services
Booking
Lodging
Apartme
nt
Hotel
Transportation
Car
Rental
service
Flight
Train
Limo
Shuttl
e
Bus
Taxi
13
Web Services Instances Selection
Filtering & Selection of Service Instances
Filters can be applied
i.e., geography
World
North
America
USA
Canada
Asia
Europe
UK
China
India
14
Travel Example Composition (complete)
Weather
Information
Intellicast.com
Address
Flight Booking
Confirmation
Arrival Date
Orbitz.com
Address (Source)
Departure
Date
Departure
time
Arrival Date
Address (Source)
Map and Driving
Direction
Location
Check- In Time
StreetMap.com
Check- Out Time
Address(Destination)
Globalhotelfinders.com
Address (hotel)
XOR
Cheapest
Hotel Reservation
Confirmation
Hotelclub.com
OUTPUT
INPUT
15
Interface-Matching Composition
Human-Assisted Composition
Web Services Composition in
Peer-to-Peer networks
16
Issues Addressed

Organization of the P2P Network



1st dimension: ontology-based
2nd dimension: interface matching-based
Dynamic Discovery of participating nodes

Based on class/subclass relationships
17
P2P organization for WS Composition
Dimension based on Web Services Ontology:
18
Conclusions and Future Work
19
Contributions

Interface-Matching composition Technique


Human-Assisted Composition Technique


Using explicit ontological service descriptions
Developed filters for easier service selection
Peer-to-Peer composition Technique

Ontology-based P2P organization
20
Future Work: Functionality-based
Composition

Functionality-based Composition


Syntax: input / output matching (WSDL level)
Semantics: Interface-Matching using ontological description

Functionality can differentiate equal semantic match

Directions:


Black-box approach:
 Exploit pre- and post-conditions in composition
White-box approach:
 State transformations (e.g., Petri nets)
21
Thanks
Questions, Comments?
22