Using context to improve data semantic
mediation in web services composition
Michaël Mrissa (spokesman) - Philippe Thiran
DBDBD’07
1
PReCISE Research Center
University of Namur, Belgium
Outline
• Introduction
–
–
–
–
Web services & composition
Semantic Web services
Mediation challenges
Objectives & contribution
• Proposition
–
–
–
–
Presentation of the context model
Context integration with Web services
General architecture
Mediation & Implementation overview
• Conclusion & perspectives
2
PReCISE Research Center
University of Namur, Belgium
Web services & composition
• Software components
• UDDI repositories
• WSDL descriptions
• SOAP messages
• Interaction model
Registry
Publish
Find
Consumer
Bind
Travel planning
Provider WS
Flight booking
• Composition of Web services
• Value-added composite Web services
Hotel booking
• Objective: answer complex user requests
• No semantics yet…
Addition
3
PReCISE Research Center
University of Namur, Belgium
Semantic Web services
• The Semantic Web
• Objectives
• Better interoperability between information systems
• Automation of information exchange
• Means
• Explicit machine-interpretable semantic descriptions
• Relies on ontologies [Gruber, 1993]
• Semantic description of Web Services
• Semantic languages
• OWL-S [Martin et al., 2004], WSMO [Arroyo and Stollberg, 2004],
DIANE [Klein et al., 2005]
• Annotation to existing formats
• WSDL : SESMA [Peer and Vukovic, 2004], WSDL-S [Miller et al., 2004]
• UDDI : [Paolucci and Kawamura, 2002]
4
PReCISE Research Center
University of Namur, Belgium
Mediation challenges
• In a composition
• Semantic mediation of exchanged data is required
Travel planning
Input
Web
Service 1
Semantic heterogeneities
Output
Input
Label conflicts
PRICE (USD)
Unit & value conflicts
PRICE (EUR)
Output
Web
Service 2
• Mediation between (semantic) Web services
• IRS-III [Cabral and Domingue, 2005],
• WSMX [Mocan et al., 2004]
• Agent-based mediation [Williams et al., 2005]
• Rule-based approach [Spencer et al., 2004]
5
PReCISE Research Center
University of Namur, Belgium
Objectives & contribution
• Multiple goals & constraints
• Solve data semantic heterogeneities in a composition
• Ease the task of Web services providers
• Explicitly describe the semantics of Web services
• Enable semantic mediation in composition
• Proposition
• Use context-based representation of data semantics
• Rely on contextual ontologies
• Annotate WSDL language with semantics
• Detect semantic heterogeneities in the composition
• Insert mediation mechanisms at the composition level
6
PReCISE Research Center
University of Namur, Belgium
Outline
• Introduction
–
–
–
–
Web services & composition
Semantic Web services
Mediation challenges
Objectives & contribution
• Proposition
–
–
–
–
Presentation of the context model
Context integration with Web services
General architecture
Mediation & Implementation overview
• Conclusion & perspectives
7
PReCISE Research Center
University of Namur, Belgium
Presentation of the context model
• Definition of “context”
• Collection of semantic assumptions on data interpretation
• How should a price be interpreted ? (VAT, currency, scale factor…)
• The context model contains 4 elements
• Semantic objects
• Static modifiers
• Dynamic modifiers
• Conversion rules & functions
• Characteristics of the model
• Good integration with WSDL
• Based on the MIX model [Bornhövd, 1999]
• Definition of static and dynamic modifiers
• Semantic conversion between semantic objects
8
PReCISE Research Center
University of Namur, Belgium
Presentation of the context model
Semantic object S = ( c, v, t, C )
• A semantic object is a 4-tuple
• A concept c defined in a domain ontology
• A value v that contains the data itself
• A type t that describes the actual type of the value
• A context C that characterizes the semantic interpretation of S
• C is described as a set of semantic objects
• Semantic object in C are called modifiers
• Modifiers are dynamic iff:
∀ v ∈ S, ∃ f:{Dom(t) ×...× Dom(t)} → Dom(t) ∧ ∃ {S1, ... Si, ... , Sn },
s.t. Si = <ci, vi, ti, Ci> ∈ Ctxt ∧ Si ≠ S ∧ f(v1, ..., vi, ..., vn) = v.
9
PReCISE Research Center
University of Namur, Belgium
Presentation of the context model
• A sample semantic object
10
PReCISE Research Center
University of Namur, Belgium
Presentation of the context model
• Conversion possibilities between semantic objects
• With context conversion functions
• Change modifiers’ values
• Dynamic aspect
• May involve access to remote resources (e.g. currency conversions)
• Stored as rules
• With type conversion functions
• Stored in conversion libraries
• Related to the type system (XML Schema)
• Semantic comparability
• Over a common type and context
• Different objects still comparable over limited context
11
PReCISE Research Center
University of Namur, Belgium
Presentation of the context model
• Illustration with the travel planning example
12
PReCISE Research Center
University of Namur, Belgium
Context integration with Web services
• Semantic annotation of WSDL metamodel
13
PReCISE Research Center
University of Namur, Belgium
Context integration with Web services
• Illustration of our annotation
• Excerpt of annotated WSDL document
• Only static modifiers are added to the description
<?xml version=``1.0" encoding=``UTF-8"?>
<wsdl:definitions...>...
<wsdl:message name=``HotelBookingTicket">
<wsdl:part name=``inputPrice" type=``xsd:double"
ctxt:context=``dom1:Price ctxt1:France
ctxt1:VATIncluded ctxt1:ScaleFactorOne"/>
</wsdl:message>...
</wsdl:definitions>
14
PReCISE Research Center
University of Namur, Belgium
Context integration with Web services
• Context ontologies
• Store context information for each domain concept
• Updated by Web services’ providers
• Separates top-down and bottom-up aspects
15
PReCISE Research Center
University of Namur, Belgium
Global architecture
16
PReCISE Research Center
University of Namur, Belgium
Mediation & Implementation overview
• Prototype
• Implementation of the travel planning example
• Graphical annotation editor (WSDL4J API)
• Development of the mediator
17
PReCISE Research Center
University of Namur, Belgium
Conclusion
• A context-based solution for semantic mediation
• A model for representing data
• Separation of domain and context knowledge
• Annotation of Web services’ descriptions
• Mediation mechanisms
• Future work
• How to insert mediators into the composition ?
• A first proposal relies on WS-BPEL analysis
• A language-independent method ?
• This model applies to Web services, but…
• What about other semantic tools on the Web ?
• Microformats, RDFa…
• To what extent the context model applies to the WWW ?
18
PReCISE Research Center
University of Namur, Belgium
Thank you !
Any questions ?
University of Namur
Contact :
http://www.fundp.ac.be
•
Email addresses
•
[email protected][email protected]
•
Authors’ web sites
•
http://www.fundp.ac.be/~mmrissa/
•
http://www.fundp.ac.be/~pthiran/
PRECISE group
http://www.fundp.ac.be/precise
19
PReCISE Research Center
University of Namur, Belgium
Descargar

Using context to improve data semantic mediation in …