CS 586 Paper Presentation

1
CS 586 Paper Presentation

• Amap A. Bhatt

2
Paper Title

• A Comparison of Two Modeling Paradigms in the
  Semantic Web
• Peter F. Patel Schneider
• Ian Horrocks

3
QUICK RECAP

• The Semantic Web?
• Information was given well defined meaning to
  form logical web of data

4
QUICK RECAP

• Ontology ?
• A Data model that represents a set of concepts
  within a domain and the relationship between
  these concepts

5
PAPER TOPIC

• A Comparison of the 2 Modeling Paradigms in the
  Semantic Web
• Classical Paradigm
• Datalog Paradigm

6
Classical Paradigm

• Based on standard Logic
• Propositional Logic
• First Order Logic
• Description Logic

7
Classical Paradigm

• Embodied in
• OWL Web Ontology Language
• RDF Resource Description Framework

8
Datalog Paradigm

• Based on Object-oriented database notions
• Embodied in
• Previous RDF version
• OWL Flight

9
DIFFERENCES

• Computational Aspects
• Expressive power
• Naturalness of Modeling

10
Which Paradigm is Better?

• Better suited for Semantic Web ?
• OWL DL vs. OWL Flight ?
• Underlying notions ?

11
Classical Paradigm

• Meaning of Information
• Mapping between Information Consistent
  Interpretation

12
Classical Paradigm

• Domain being Modeled Abstract representation
  of Objects and relationships between them

13
Classical Paradigm

• Information and Interpretation as separate
  Entities.
• More than 1 interpretation possible.
• Query checking some situation that holds in
  all interpretations

14
Classical Paradigm

• Use of Axioms
• Example. Tim Pam ? USC

15
Datalog Paradigm

• Closed World Assumption
• Unique Name Assumption
• Result gt Single Model (1 to 1)

16
Datalog Paradigm

• Formalization of the DB Approach
• Horn like rules to capture Schema Data
• Person(y) ? Person(x) Parent(x,y)

17
Datalog Paradigm

• Query checking ? checking the single Model
• Tradeoff Ease vs. Scope

18
Datalog Paradigm Tradeoff

• Polynomial time Query answering
• All persons are either male or female
• All persons have exactly 2 parents

19
Identifiers Open World

• Identifier uniqueness
• Epistemic queries and constructs.

20
Incomplete Information

• John Smith knows 2 people
• D.O.B P.O.B for every person
• Person(John_Smith)

21
Constraints

• Adding new Information
• Datalog limitations

22
Constraints

• Rule
• Mother(x,y)? Person(x)
• Constraint Rule
• ?Person(x)Parent (x,x)

23
Datatypes

• Datatypes Classes
• Datatype values Identifiers
• 1 lt 2
• 1 and 1.0

24
Role of Tools

• Ontology Building Tools with Good User Interfaces
• Characteristics of good Tool?

25
Complexity

• Classical vs. Datalog
• Datalog Tradeoff
• Scope vs. Reasoning Time

26
Integrating DL and Rules

• Safety Condition
• Impact on Complexity

27
Extending Classical Paradigm

• Local Closed / Unique name assumption by using
  Constructs
• Extensions to OWL DL

28
Discussion Issues

• Inherent openness of Semantic Web
• Solution Ontology Tools.
• (add epistemic constructs to OWL)
• Right Balance