[ISDA'11] Towards integrating fuzzy logic capabilities into an ontology based inductive logic programming framework
1. Grupo de Procesado de Datos y Simulación
ETSI de Telecomunicación
Universidad Politécnica de Madrid
Towards Integrating Fuzzy Logic Capabilities into an
Ontology-based Inductive Logic Programming Framework
ISDA 2011
Josué Iglesias, Jens Lehmann
josue@grpss.ssr.upm.es
2. contents
DL-based Inductive Logic Programming
DL-Learner overview
extending DL-Learner with fuzzy support
semantic ILP problem test case
experiments & results
conclusions
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 1 / 10
3. contents
DL-based Inductive Logic Programming
DL-Learner overview
extending DL-Learner with fuzzy support
semantic ILP problem test case
experiments & results
conclusions
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 1 / 10
4. DL-based Inductive Logic Programming
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 1 / 10
5. DL-based Inductive Logic Programming
OWL-DL ontology
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 1 / 10
6. DL-based Inductive Logic Programming
OWL-DL ontology DL Knowledge Base
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 1 / 10
7. DL-based Inductive Logic Programming
OWL-DL ontology DL Knowledge Base +/- examples
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 1 / 10
8. DL-based Inductive Logic Programming
OWL-DL ontology DL Knowledge Base +/- examples
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 1 / 10
9. DL-based Inductive Logic Programming
OWL-DL ontology DL Knowledge Base +/- examples
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 1 / 10
10. DL-based Inductive Logic Programming
OWL-DL ontology DL Knowledge Base +/- examples
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 1 / 10
11. DL-based Inductive Logic Programming
OWL-DL ontology DL Knowledge Base +/- examples
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 1 / 10
12. contents
DL-based Inductive Logic Programming
DL-Learner overview
extending DL-Learner with fuzzy support
semantic ILP problem test case
experiments & results
conclusions
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 2 / 10
13. DL-Learner overview
+ +/-
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 2 / 10
14. DL-Learner overview
+ +/-
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 2 / 10
15. DL-Learner overview
+ +/-
solutions
#1 [100%]
#2 [99.43%]
...
#N [83.12%]
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 2 / 10
16. DL-Learner overview
+ +/-
solutions
#1 [100%]
#2 [99.43%]
...
#N [83.12%]
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 2 / 10
17. DL-Learner overview
+ +/-
fuzzy
support?? solutions
#1 [100%]
#2 [99.43%]
...
#N [83.12%]
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 2 / 10
18. contents
DL-based Inductive Logic Programming
DL-Learner overview
extending DL-Learner with fuzzy support
semantic ILP problem test case
experiments & results
conclusions
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 3 / 10
19. extending DL-Learner with fuzzy support
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 3 / 10
20. extending DL-Learner with fuzzy support
fuzzy knowledge source component
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 3 / 10
21. extending DL-Learner with fuzzy support
fuzzy knowledge source component
objective
• classical (crisp) ontologies/KB fuzzy ontologies/KB support
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 3 / 10
22. extending DL-Learner with fuzzy support
fuzzy knowledge source component
objective
• classical (crisp) ontologies/KB fuzzy ontologies/KB support
candidates
• meta-ontologies
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 3 / 10
23. extending DL-Learner with fuzzy support
fuzzy knowledge source component
objective
• classical (crisp) ontologies/KB fuzzy ontologies/KB support
candidates
• meta-ontologies
• OWL recommendation extensions
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 3 / 10
24. extending DL-Learner with fuzzy support
fuzzy knowledge source component
objective
• classical (crisp) ontologies/KB fuzzy ontologies/KB support
candidates
• meta-ontologies
• OWL recommendation extensions
• OWL-based available tools
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 3 / 10
25. extending DL-Learner with fuzzy support
fuzzy knowledge source component
objective
• classical (crisp) ontologies/KB fuzzy ontologies/KB support
candidates
• meta-ontologies
semantic-friendly
less human-understandable
• OWL recommendation extensions
• OWL-based available tools
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 3 / 10
26. extending DL-Learner with fuzzy support
fuzzy knowledge source component
objective
• classical (crisp) ontologies/KB fuzzy ontologies/KB support
candidates
• meta-ontologies
semantic-friendly
less human-understandable
• OWL recommendation extensions
solid semantics
no standardization expected
• OWL-based available tools
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 3 / 10
27. extending DL-Learner with fuzzy support
fuzzy knowledge source component
objective
• classical (crisp) ontologies/KB fuzzy ontologies/KB support
candidates
• meta-ontologies
semantic-friendly
less human-understandable
• OWL recommendation extensions
solid semantics
no standardization expected
• OWL-based available tools
standard & tools compatible
sematic-less approach
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 3 / 10
28. extending DL-Learner with fuzzy support
fuzzy knowledge source component
objective
• classical (crisp) ontologies/KB fuzzy ontologies/KB support
candidates
• meta-ontologies
semantic-friendly
less human-understandable
• OWL recommendation extensions
solid semantics
no standardization expected
• OWL-based available tools
standard & tools compatible
sematic-less approach
final implementation
• FuzzyOWL2 (fuzzy OWL annotations)
active project
parser FuzzyOWL2 reasoners syntax
OWLAPI
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 3 / 10
29. extending DL-Learner with fuzzy support
fuzzy knowledge source component
objective
• classical (crisp) ontologies/KB fuzzy ontologies/KB support
candidates
• meta-ontologies
semantic-friendly
less human-understandable
• OWL recommendation extensions
solid semantics
no standardization expected
• OWL-based available tools
standard & tools compatible
sematic-less approach
final implementation
• FuzzyOWL2 (fuzzy OWL annotations)
active project
parser FuzzyOWL2 reasoners syntax
OWLAPI
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 3 / 10
30. extending DL-Learner with fuzzy support
fuzzy knowledge source component
objective
• classical (crisp) ontologies/KB fuzzy ontologies/KB support
candidates
• meta-ontologies
semantic-friendly
less human-understandable
• OWL recommendation extensions
solid semantics
no standardization expected
• OWL-based available tools
standard & tools compatible
sematic-less approach
final implementation
• FuzzyOWL2 (fuzzy OWL annotations) fuzzy classes definitions
fuzzy concept assertions
active project fuzzy data types definitions
parser FuzzyOWL2 reasoners syntax fuzzy modifiers
fuzzy role assertions
OWLAPI etc.
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 3 / 10
31. extending DL-Learner with fuzzy support
fuzzy knowledge source component
objective
• classical (crisp) ontologies/KB fuzzy ontologies/KB support
candidates
• meta-ontologies
semantic-friendly
less human-understandable Protégé plug-in
• OWL recommendation extensions
solid semantics
no standardization expected
• OWL-based available tools
standard & tools compatible
sematic-less approach
final implementation
• FuzzyOWL2 (fuzzy OWL annotations) fuzzy classes definitions
fuzzy concept assertions
active project fuzzy data types definitions
parser FuzzyOWL2 reasoners syntax fuzzy modifiers
fuzzy role assertions
OWLAPI etc.
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 3 / 10
32. extending DL-Learner with fuzzy support
fuzzy reasoning service component
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 4 / 10
33. extending DL-Learner with fuzzy support
fuzzy reasoning service component
objective
• support fuzzy ontology/KB reasoning
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 4 / 10
34. extending DL-Learner with fuzzy support
fuzzy reasoning service component
objective
• support fuzzy ontology/KB reasoning
candidates
• few candidates
[based on reduction procedures (e.g., DeLorean)]
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 4 / 10
35. extending DL-Learner with fuzzy support
fuzzy reasoning service component
objective
• support fuzzy ontology/KB reasoning
candidates
• few candidates
[based on reduction procedures (e.g., DeLorean)]
GERDS GURDL FiRE Yadlr fuzzyDL
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 4 / 10
36. extending DL-Learner with fuzzy support
fuzzy reasoning service component
objective
• support fuzzy ontology/KB reasoning
candidates
• few candidates
[based on reduction procedures (e.g., DeLorean)]
GERDS GURDL FiRE Yadlr fuzzyDL
• availability
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 4 / 10
37. extending DL-Learner with fuzzy support
fuzzy reasoning service component
objective
• support fuzzy ontology/KB reasoning
candidates
• few candidates
[based on reduction procedures (e.g., DeLorean)]
GERDS GURDL FiRE Yadlr fuzzyDL
• availability
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 4 / 10
38. extending DL-Learner with fuzzy support
fuzzy reasoning service component
objective
• support fuzzy ontology/KB reasoning
candidates
• few candidates
[based on reduction procedures (e.g., DeLorean)]
GERDS GURDL FiRE Yadlr fuzzyDL
• availability
• Java-based interface (to ease DL-Learner integration)
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 4 / 10
39. extending DL-Learner with fuzzy support
fuzzy reasoning service component
objective
• support fuzzy ontology/KB reasoning
candidates
• few candidates
[based on reduction procedures (e.g., DeLorean)]
GERDS GURDL FiRE Yadlr fuzzyDL
• availability
• Java-based interface (to ease DL-Learner integration)
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 4 / 10
40. extending DL-Learner with fuzzy support
fuzzy reasoning service component
objective
• support fuzzy ontology/KB reasoning
candidates
• few candidates
[based on reduction procedures (e.g., DeLorean)]
GERDS GURDL FiRE Yadlr fuzzyDL
• availability
• Java-based interface (to ease DL-Learner integration)
• FuzzyOWL2 integration
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 4 / 10
41. extending DL-Learner with fuzzy support
fuzzy reasoning service component
objective
• support fuzzy ontology/KB reasoning
candidates
• few candidates
[based on reduction procedures (e.g., DeLorean)]
GERDS GURDL FiRE Yadlr fuzzyDL
• availability
• Java-based interface (to ease DL-Learner integration)
• FuzzyOWL2 integration
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 4 / 10
42. extending DL-Learner with fuzzy support
fuzzy reasoning service component
objective
• support fuzzy ontology/KB reasoning
candidates
• few candidates
[based on reduction procedures (e.g., DeLorean)]
GERDS GURDL FiRE Yadlr fuzzyDL
• availability
• Java-based interface (to ease DL-Learner integration)
• FuzzyOWL2 integration
final implementation
• fuzzyDL
active project (most up-to-date development)
Java-based
parser FuzzyOWL2 fuzzyDL reasoner syntax
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 4 / 10
43. extending DL-Learner with fuzzy support
fuzzy reasoning service component
objective
• support fuzzy ontology/KB reasoning
candidates
• few candidates
[based on reduction procedures (e.g., DeLorean)]
GERDS GURDL FiRE Yadlr fuzzyDL
• availability
• Java-based interface (to ease DL-Learner integration)
• FuzzyOWL2 integration
final implementation
• fuzzyDL
active project (most up-to-date development) OWLAPI
Java-based encapsulation
parser FuzzyOWL2 fuzzyDL reasoner syntax (and extension)
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 4 / 10
44. extending DL-Learner with fuzzy support
fuzzy reasoning service component
GERDS GURDL FiRE Yadlr fuzzyDL
OWLAPI
encapsulation
(and extension)
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 4 / 10
45. extending DL-Learner with fuzzy support
fuzzy learning problem component
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 5 / 10
46. extending DL-Learner with fuzzy support
fuzzy learning problem component
objective
• +/- examples
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 5 / 10
47. extending DL-Learner with fuzzy support
fuzzy learning problem component
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 5 / 10
48. extending DL-Learner with fuzzy support
fuzzy learning problem component
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 5 / 10
49. extending DL-Learner with fuzzy support
fuzzy learning problem component
fuzzy F-measure
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 5 / 10
50. extending DL-Learner with fuzzy support
fuzzy learning problem component
fuzzy F-measure
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 5 / 10
51. extending DL-Learner with fuzzy support
fuzzy learning problem component
fuzzy F-measure
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 5 / 10
52. extending DL-Learner with fuzzy support
fuzzy learning problem component
fuzzy F-measure
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 5 / 10
53. extending DL-Learner with fuzzy support
fuzzy learning problem component
fuzzy F-measure
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 5 / 10
54. extending DL-Learner with fuzzy support
fuzzy learning problem component
solutions
#1 [100%]
#2 [99.43%]
...
#N [83.12%]
fuzzy F-measure
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 5 / 10
55. extending DL-Learner with fuzzy support
fuzzy learning algorithm component
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 6 / 10
56. extending DL-Learner with fuzzy support
fuzzy learning algorithm component
algorithm
• candidate concepts generation
• CELOE (Class Expression Learning for Ontology Engineering)
others already developed and ready to be used
(refinement operator techniques)
• no significant modifications
fuzzy reasoner OWLAPI encapsulation (and extension)
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 6 / 10
57. contents
DL-based Inductive Logic Programming
DL-Learner overview
extending DL-Learner with fuzzy support
semantic ILP problem test case
experiments & results
conclusions
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 7 / 10
58. semantic ILP problem test case
Michalski’s train ILP problem [16]
[16] J. Larson and R. S. Michalski, “Inductive inference of VL decision rules,” SIGART Bull., pp. 38–44, June 1977.
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 7 / 10
59. semantic ILP problem test case
Michalski’s train ILP problem [16] Konstantopoulos–Charalambidis’
fuzzy trains ILP problem [3]
fuzzy definition
fuzzy reasoning
[3] S. Konstantopoulos and A. Charalambidis, “Formulating description logic learning as an inductive logic programming task,” in
Proc. of FUZZ-IEEE, 2010 IEEE World Congress on Comp. Int., July 18–23, Barcelona. IEEE, Jul. 2010.
[16] J. Larson and R. S. Michalski, “Inductive inference of VL decision rules,” SIGART Bull., pp. 38–44, June 1977.
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 7 / 10
60. semantic ILP problem test case
Michalski’s train ILP problem [16] Konstantopoulos–Charalambidis’
fuzzy trains ILP problem [3]
fuzzy definition OWL-based
fuzzy reasoning generic tool
[3] S. Konstantopoulos and A. Charalambidis, “Formulating description logic learning as an inductive logic programming task,” in
Proc. of FUZZ-IEEE, 2010 IEEE World Congress on Comp. Int., July 18–23, Barcelona. IEEE, Jul. 2010.
[16] J. Larson and R. S. Michalski, “Inductive inference of VL decision rules,” SIGART Bull., pp. 38–44, June 1977.
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 7 / 10
61. semantic ILP problem test case
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 8 / 10
62. semantic ILP problem test case
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 8 / 10
63. semantic ILP problem test case
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 8 / 10
64. semantic ILP problem test case
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 8 / 10
65. semantic ILP problem test case
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 8 / 10
66. semantic ILP problem test case
fuzzy class definition
fuzzy concept assertion
fuzzy role assertion
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 8 / 10
67. contents
DL-based Inductive Logic Programming
DL-Learner overview
extending DL-Learner with fuzzy support
semantic ILP problem test case
experiments & results
conclusions
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 9 / 10
69. experiments & results
crisp trains KB
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 9 / 10
70. experiments & results
OWL2
FuzzyOWL2
crisp trains KB
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 9 / 10
71. experiments & results
OWL2
FuzzyOWL2 + fuzzy
crisp trains KB
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 9 / 10
72. experiments & results
OWL2
=?
FuzzyOWL2 + fuzzy
crisp trains KB
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 9 / 10
73. experiments & results
OWL2
=?
FuzzyOWL2 + fuzzy
crisp trains KB
FuzzyOWL2 + fuzzy
fuzzy trains KB
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 9 / 10
74. experiments & results
OWL2
=?
FuzzyOWL2 + fuzzy
crisp trains KB
FuzzyOWL2 + fuzzy
fuzzy trains KB
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 9 / 10
75. experiments & results
OWL2
=?
FuzzyOWL2 + fuzzy
crisp trains KB
FuzzyOWL2 + fuzzy
fuzzy trains KB
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 9 / 10
76. contents
DL-based Inductive Logic Programming
DL-Learner overview
extending DL-Learner with fuzzy support
semantic ILP problem test case
experiments & results
conclusions
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 10 / 10
77. conclusions
• DL-Learner + fuzzy support
general purpose ILP tool
OWL-based
open-source
http://dl-learner.org
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 10 / 10
78. conclusions
• DL-Learner + fuzzy support
general purpose ILP tool
OWL-based
open-source
http://dl-learner.org
• real integration of up-to-date fuzzy ontologies tools
FuzzyOWL2: fuzzy ontology representation tool
fuzzyDL: fuzzy DL reasoner
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 10 / 10
79. conclusions
• DL-Learner + fuzzy support
general purpose ILP tool
OWL-based
open-source
http://dl-learner.org
• real integration of up-to-date fuzzy ontologies tools
FuzzyOWL2: fuzzy ontology representation tool
fuzzyDL: fuzzy DL reasoner
• preliminar evaluation
functional correctness
response time ( OWLAPI encapsulation new fuzzy reasoners)
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 10 / 10
80. conclusions
• DL-Learner + fuzzy support
general purpose ILP tool
OWL-based
open-source
http://dl-learner.org
• real integration of up-to-date fuzzy ontologies tools
FuzzyOWL2: fuzzy ontology representation tool
fuzzyDL: fuzzy DL reasoner
• preliminar evaluation
functional correctness
response time ( OWLAPI encapsulation new fuzzy reasoners)
extend test case
more complex fuzzy trains example
real world problem (recommendation / shop assistant)
International Conference on Intelligent Systems Design and Applications josue@grpss.ssr.upm.es 10 / 10