Empowering OWL with Overriding Inheritance, Conflict Resolution and Non-monotonic Reasoning

1
Empowering OWL with Overriding Inheritance,
Conflict Resolution andNon-monotonic Reasoning

• Presented By
• Md. Shazzad Hosain

2
6/5/2020
3
OWL Web Ontology Language

• Semantic Web requires ontology language to reason
  about concepts and relationships
• RDF, OWL, OIL, DAMLOIL, FLORA-2 etc.
• W3C accepted OWL as standard in 2004.
• Inherent limitation in representing Defaults
  and/or exceptions

6/5/2020
4
An Intuitive Example
monthlyFee
monthlyFee 100
Donor
Donor
100
100
200
100
Joe
Max
John
Joe
Max
John
6/5/2020
5
OWL doesnt handle defaults

• Usually more expressive language are more
  computationally expensive
• Sometimes decidability of languages suffer
• OWL chooses to be decidable and computationally
  efficient i.e. unlike first order theorem
  provers, such as Prolog, the reasoning process
  should always terminate
• OWL has limited expressivity in terms of
  exception handling

6/5/2020
6
But Default is important for KR

• Bird can fly, penguin is a bird but cant fly
• Rules with exception Give discount to loyal
  customer, but not discount may be allowed to
  customers
• Reasoning with incomplete information
  conventional wisdom dictates that packaging for
  tax free food is also tax free. But, any
  packaging that is more exotic than usual is not
  tax free.
• Similarly, while ontology merging exception could
  occur

6/5/2020
7
Application development perspective

• C, Java are object oriented.
• Semantic applications requires object-ontology
  mapping, like Object-relational mapping
• This is known as impedance mismatch
• Many frameworks are being developed for automatic
  mapping between these two paradigms

6/5/2020
8
Our Goal

• Extend OWL to a more expressive language called
  OWL
• Include Default Inheritance
• Conflict resolution in multiple inheritance
• Null inheritance
• Maintain strengths of OWL
• Reduce impedance mismatch

6/5/2020
9
Motivating Example
Donor
Donor
Donor
Social Net System of Non Profit Organization
Mentor
Mentor
Student
Student
Student
Student
Student
6/5/2020
10
Motivating Example
6/5/2020
11
OWL Syntax
monthlyFee
Donor
100
100
100
Joe
Max
John
6/5/2020
12
OWL Syntax
monthly_fee 100
Donor
monthly_fee 200
Joe
Max
Jhon
6/5/2020
13
(No Transcript)
14
(No Transcript)
15
(No Transcript)
16
(No Transcript)
17
OWL Semantics

• We give translational semantics to OWL
• We transform every OWL program to an equivalent
  OWL program and
• set of rules
• Preserves OWLs strengths
• Use existing OWL reasoners to
• build OWL reasoner
• Two steps of translation
• Compute Inheritability function
• Rewrite OWL to OWL

6/5/2020
18
Inheritable (Donor, monthlyFee,
Donor) Inheritable (Donor, type,
Donor) Inheritable (Mentor, profession,
Mentor) Inheritable (Associate, profession,
Mentor) Inheritable (Associate, monthlyFee,
Donor) Inheritable (Associate, type, Donor)
19
Rewriting OWL Classes
monthlyFee 100
Donor
monthlyFee 200
Joe
Max
John
OWL Structure
Equivalent OWL Structure
6/5/2020
19
20
6/5/2020
21
Rewriting Queries
QOWL

• We use SPARQL to query OWL
• Syntax (subject predicate object)
• Example, (D1 monthlyFee ?value)
• For flexibility we will use monthlyFee (D1,
  value)
• Class queries like monthlyFee(Donor, value)
• property(source, value) lt- inheritable(source,
  prop, object), property (object, value)
• property(souce, value) lt- inheritable(souce,
  prop, object), sysInstanceOf(object, sys_object),
  property(sys_object, value)

6/5/2020
22
Conclusion

• Extend OWL to OWL with default inheritance,
  conflict resolution, null inheritance
• We have shown translational semantics of OWL
• Thus build OWL reasoner using existing OWL
  reasoner Pellete and Jena rule engine

6/5/2020
22
23
Thanks