A Semantic Web Service Discovery and Composition Prototype Framework Using Production Rules

1
A Semantic Web Service Discovery and Composition
Prototype Framework Using Production Rules

• Georgios MeditskosNick Bassiliades
• Aristotle University of Thessaloniki, Greece

Intelligent Systems and Knowledge Processing group
1
2
Outline

• Motivation
• Why OWL-S ?
• Benefits of Rules and Object-Oriented Paradigm
• Discovery
• Composition
• Open Issues

3
Motivation

• Handle Semantic Web Services using a rule engine
• robustness, declarativeness
• Investigate the usability of the Object-Oriented
  representation on
• OWL reasoning
• Matchmaking
• Composition
• modularity, indexing

OWL-S
4
ProSeDisCo

• Production rules for Service Discovery and
  Composition
• The framework combines
• Production rules for OWL(-S) reasoning
• Object-oriented representation of ontologies
• Underlying system O-DEVICE (OO Rule-based OWL
  reasoner)

5
Outline

• Motivation
• Why OWL-S ?
• Benefits of Rules and Object-Oriented Paradigm
• Discovery
• Composition
• Open Issues

6
Why OWL-S?

• Quick implementation of a prototype
• Existing infrastructure for OWL reasoning
• Loading the ontology into O-DEVICE
• Implementing production rules to facilitate
  discovery composition
• WSMO introduces a more complicated conceptual
  model
• OWL-S is a more mature approach

7
Outline

• Motivation
• Why OWL-S ?
• Benefits of Rules and Object-Oriented Paradigm
• Discovery
• Composition
• Open Issues

8
Object-Oriented Rules for OWL

• A transformation procedure of OWL into the OO
  model (CLIPS/COOL)
• A rule-based approach transforms
• OWL classes into COOL classes
• OWL properties into class slots
• OWL instances into COOL objects
• Object-Oriented entailments
• Rules operate over the OO schema

9
Why Rules ?

• Simplistic model for knowledge representation
• Domain experts
• Programmers
• Ontologies with rule programs
• Flexible and declarative programming
• Rule programming based on OWL semantics

10
Advantages of the OO Approach

• Acts as an indexing technique
• Exploits the message passing mechanism
• (send george get-hasBrother)
• Single multiple inheritance is treated by the
  OO environment
• Quick determination of subsumption relations
• Built-in functions for determining hierarchical
  relationships
• Can be implemented in any OO environment

11
OWL-S and OO Model

• Web service capabilities are defined as instances
  of the Profile class
• Each profile is mapped into the corresponding OO
  Profile class
• Result An OO KB with object advertisements
• Requests are also Profile objects
• Matchmaking Determine similar objects

12
Outline

• Motivation
• Why OWL-S ?
• Benefits of Rules and Object-Oriented Paradigm
• Discovery
• Composition
• Open Issues

13
Semantic Discovery

• Goal
• Determine Web services that satisfy certain
  requirements
• Methodology
• Check input/output parameters
• Compute datatype or class relationships

14
Datatypes

• Direct comparison of datatypes
• Three types of match
• Exact (e.g. xsdint, xsdint)
• Numerical (e.g. xsdint, xsdfloat)
• Mismatch (e.g. xsddouble, xsdboolean)
• Different weights for each match
• exact gt numerical gt mismatch

15
Classes

• Examine class hierarchy
• Subsumption relations are determined over the OO
  schema
• Using of built-in CLIPS functions
• subclass(A, B) -gt TRUE
• superclasses(A) -gt
• Advantage
• Quick determination

16
Three types of class match

• Exact
• The same class
• Equivalent class
• Plug-in or subsume
• Hierarchical relationship
• Fail
• There is not hierarchical relationship
• Disjoint classes
• Matchmaking Checking advertisements inputs
  against a request inputs and request outputs
  against advertisements outputs

17
Outline

• Motivation
• Why OWL-S ?
• Benefits of Rules and Object-Oriented Paradigm
• Discovery
• Composition
• Open Issues

18
Composition

• The algorithm is implemented with production
  rules
• Traverses Profile objects
• Finds appropriate objects (advertisements) that
  satisfy requirements at each step
• More than one composition plans
• Scores each plan based only on I/O matches

19
Composition Procedure

1. Selects Web services that satisfy all or some
   request outputs
2. Creates all the combination sets of Web services
   that satisfy all request outputs
3. Recursive algorithm in order to satisfy each
   sets inputs
4. The algorithm terminates when all the inputs of
   all Web services of a specific set are satisfied

20
Example

• Request(inTitle, outPrice, Publisher)

• WS1(inTitle, outPublisher)
• WS2(inTitle, outPerson)
• WS3(inTitle, outISBN)
• WS4(inISBN, outPrice)

21
Example (cont.)

• Two initial sets of Web services

22
Example (cont.)
Request(inTitle, outPrice, Publisher)
23
Outline

• Motivation
• Why OWL-S ?
• Benefits of Rules and Object-Oriented Paradigm
• Discovery
• Composition
• Open Issues

24
Open Issues

• Non-functional properties
• Web service discovery based on non-functional
  properties
• Classification
• Classification of Web services based on their
  functionality
• Filtering of Web services during discovery
• Preconditions Effects
• Filtering of Web services
• Mapped into production rules

25
Open Issues

• Process Model
• Both on discovery and invocation
• Current Work
• Exploit more hierarchical relations stemming from
  the OO model on discovery
• sibling classes

26
Thank you for your attention