+ All Categories
Home > Technology > A GUI for MLN

A GUI for MLN

Date post: 07-May-2015
Category:
Upload: rommel-carvalho
View: 276 times
Download: 4 times
Share this document with a friend
Description:
Presentation given by Rommel N. Carvalho at the 9th International Workshop on Uncertainty Reasoning for the Semantic Web at the 12th International Semantic Web Conference in October 21, 2013, Sydney, Australia. This was a joint work between the Research and Strategic Information Directorate from Brazil's Office of the Comptroller General and the Department of Computer Science from the University of Brasília. Title: A GUI for MLN. Abstract: This paper focuses on the incorporation of the Markov Logic Network (MLN) formalism as a plug-in for UnBBayes, a Java framework for probabilistic reasoning based on graphical models. MLN is a formalism for probabilistic reasoning which combines the capacity of dealing with uncertainty tolerating imperfections and contradictory knowledge based a Markov Network (MN) with the expressiveness of First Order Logic. A MLN provides a compact language for specifying very large MNs and the ability to incorporate, in modular form, large domain of knowledge (expressed in First Order Logic sentences) inside itself. A Graphical User Interface for the software Tuffy was implemented into UnBBayes to facilitate the creation, and inference of MLN models. Tuffy is a Java open source MLN engine.
38
A GUI for MLN Estêvão F. Aguiar, Marcelo Ladeira, Rommel N. Carvalho, and Shou Matsumoto Universidade de Brasília Paper - Uncertainty Reasoning for the Semantic Web URSW - ISWC 10/21/2013 - Sydney, Australia
Transcript
Page 1: A GUI for MLN

A GUI for MLNEstêvão F. Aguiar, Marcelo Ladeira, Rommel N. Carvalho, and Shou

Matsumoto Universidade de Brasília

!Paper - Uncertainty Reasoning for the Semantic Web

URSW - ISWC 10/21/2013 - Sydney, Australia

Page 2: A GUI for MLN

Agenda

2

Page 3: A GUI for MLN

Agenda

Markov Logic Network (MLN)

2

Page 4: A GUI for MLN

Agenda

Markov Logic Network (MLN)

Choice of Implementation

2

Page 5: A GUI for MLN

Agenda

Markov Logic Network (MLN)

Choice of Implementation

To GUI or Not to GUI

2

Page 6: A GUI for MLN

Agenda

Markov Logic Network (MLN)

Choice of Implementation

To GUI or Not to GUI

Conclusion

2

Page 7: A GUI for MLN

MLN

3MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 8: A GUI for MLN

MLN*

4

* [Richardson & Domingos 2006] http://hazy.cs.wisc.edu/hazy/papers/tuffy-vldb2011-slides.pdf

MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 9: A GUI for MLN

MLN*Syntax: a set of weighted logical rules

0.5 Smokes(p) => Cancer(p) // if you smoke, you are more likely to have cancer

 Weights: cost for rule violation

4

* [Richardson & Domingos 2006] http://hazy.cs.wisc.edu/hazy/papers/tuffy-vldb2011-slides.pdf

MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 10: A GUI for MLN

MLN*Syntax: a set of weighted logical rules

0.5 Smokes(p) => Cancer(p) // if you smoke, you are more likely to have cancer

 Weights: cost for rule violation

Semantics: a distribution over possible worlds

Each possible world I incurs total cost cost(I)

Pr[I] ∝exp(−cost(I))

Thus most likely world has lowest cost

4

* [Richardson & Domingos 2006] http://hazy.cs.wisc.edu/hazy/papers/tuffy-vldb2011-slides.pdf

MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 11: A GUI for MLN

MLN*

5

* http://hazy.cs.wisc.edu/hazy/papers/tuffy-vldb2011-slides.pdf

MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 12: A GUI for MLN

MLN*

5

Rules!

0.5 Smokes(p) => Cancer(p) // if you smoke, you are more likely to have cancer

0.4 Friends(p1,p2) ∧ Smokes(p1) => Smokes(p2) 0.4 Friends(p1,p2) ∧ Smokes(p2) => Smokes(p1)

// friends that smoke are more likely to have friends that also smoke

* http://hazy.cs.wisc.edu/hazy/papers/tuffy-vldb2011-slides.pdf

MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 13: A GUI for MLN

MLN*

5

Rules!

0.5 Smokes(p) => Cancer(p) // if you smoke, you are more likely to have cancer

0.4 Friends(p1,p2) ∧ Smokes(p1) => Smokes(p2) 0.4 Friends(p1,p2) ∧ Smokes(p2) => Smokes(p1)

// friends that smoke are more likely to have friends that also smoke

* http://hazy.cs.wisc.edu/hazy/papers/tuffy-vldb2011-slides.pdf

Certain!

Friends(Anna,Bob) Smokes(Anna)

MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 14: A GUI for MLN

MLN*

5

Rules!

0.5 Smokes(p) => Cancer(p) // if you smoke, you are more likely to have cancer

0.4 Friends(p1,p2) ∧ Smokes(p1) => Smokes(p2) 0.4 Friends(p1,p2) ∧ Smokes(p2) => Smokes(p1)

// friends that smoke are more likely to have friends that also smoke

* http://hazy.cs.wisc.edu/hazy/papers/tuffy-vldb2011-slides.pdf

Certain!

Friends(Anna,Bob) Smokes(Anna)

Uncertain!

Cancer(?)

MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 15: A GUI for MLN

Choice of Implementation

6MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 16: A GUI for MLN

UnBBayes Plugin Framework

7MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 17: A GUI for MLN

Choice of Implementation

8MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 18: A GUI for MLN

To GUI or Not to GUI

9MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 19: A GUI for MLN

To GUI or Not to GUI*

10

*http://hazy.cs.wisc.edu/hazy/tuffy/doc/

MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 20: A GUI for MLN

To GUI or Not to GUI*

11

*http://hazy.cs.wisc.edu/hazy/tuffy/doc/

MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 21: A GUI for MLN

To GUI or Not to GUI*

11

*http://hazy.cs.wisc.edu/hazy/tuffy/doc/

MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 22: A GUI for MLN

To GUI or Not to GUI*

11

*http://hazy.cs.wisc.edu/hazy/tuffy/doc/

MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 23: A GUI for MLN

To GUI or Not to GUI*

11

*http://hazy.cs.wisc.edu/hazy/tuffy/doc/

MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 24: A GUI for MLN

To GUI or Not to GUI*

11

*http://hazy.cs.wisc.edu/hazy/tuffy/doc/

MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 25: A GUI for MLN

To GUI or Not to GUI*

11

*http://hazy.cs.wisc.edu/hazy/tuffy/doc/

MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 26: A GUI for MLN

To GUI or Not to GUI

12MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 27: A GUI for MLN

To GUI or Not to GUI

13MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 28: A GUI for MLN

To GUI or Not to GUI

14MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 29: A GUI for MLN

To GUI or Not to GUI

15MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 30: A GUI for MLN

To GUI or Not to GUI

16MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 31: A GUI for MLN

To GUI or Not to GUI

16MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 32: A GUI for MLN

To GUI or Not to GUI

16MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 33: A GUI for MLN

To GUI or Not to GUI

17MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 34: A GUI for MLN

Conclusion

18MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 35: A GUI for MLN

Conclusion

19MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 36: A GUI for MLN

ConclusionPros

Easy-to-use tool for MLN

Dynamic search of predicates, rules, and evidence

Add/remove features for predicates, rules, and evidence

Facilitates new parameters available in future versions of Tuffy

19MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 37: A GUI for MLN

ConclusionPros

Easy-to-use tool for MLN

Dynamic search of predicates, rules, and evidence

Add/remove features for predicates, rules, and evidence

Facilitates new parameters available in future versions of Tuffy

Cons

Still a beta tool

Needs more serious testing

A lot more GUI features could be incorporated

19MLN - Choice of Implementation - To GUI or Not to GUI - Conclusion

Page 38: A GUI for MLN

Obrigado!

20


Recommended