Graphical Models and Probabilistic Reasoning for Generating Linked Data from Tables

1
Graphical Models and Probabilistic Reasoning for
Generating Linked Data from Tables

• Varish Mulwad (_at_varish)University of Maryland,
  Baltimore CountyDoctoral Consortium at ISWC
  2011October 24, 2011

Guru Dr. Tim Finin
2
What ?
3
Contribution
http//dbpedia.org/class/yago/NationalBasketballAs
sociationTeams
dbpropteam
Name Team Position Height
Michael Jordan Chicago Shooting guard 1.98
Allen Iverson Philadelphia Point guard 1.83
Yao Ming Houston Center 2.29
Tim Duncan San Antonio Power forward 2.11
http//dbpedia.org/resource/Allen_Iverson
Map literals as values of properties
4
Contribution
Name Team Position Height
Michael Jordan Chicago Shooting guard 1.98
Allen Iverson Philadelphia Point guard 1.83
Yao Ming Houston Center 2.29
Tim Duncan San Antonio Power forward 2.11
_at_prefix dbpedia lthttp//dbpedia.org/resource/gt
. _at_prefix dbpedia-owl lthttp//dbpedia.org/ontolog
y/gt . _at_prefix yago lthttp//dbpedia.org/class/yago
/gt . "Name"_at_en is rdfslabel of
dbpedia-owlBasketballPlayer . "Team"_at_en is
rdfslabel of yagoNationalBasketballAssociationTe
ams . "Michael Jordan"_at_en is rdfslabel of
dbpediaMichael Jordan . dbpediaMichael Jordan a
dbpedia-owlBasketballPlayer . "Chicago
Bulls"_at_en is rdfslabel of dbpediaChicago Bulls
. dbpediaChicago Bulls a yagoNationalBasketballA
ssociationTeams .
All this in a completely automated way !! ?
5
Why ?
6
Tables are everywhere !! yet
The web 154 million high quality relational
tables 1
389, 697 raw and geospatial datasets0.071 in
RDF
7
Current Systems

• Problems with systems on the Semantic Web
• Require users to have knowledge of the Semantic
  Web
• Do not automatically link to existing classes and
  entities on the Semantic Web / Linked Data cloud
• RDF data in some cases is as useless as raw data
• Majority of the work focused on relational data
  where schema is available

8
How ?
9
A Table Interpretation Framework
Linked Data
Probabilistic Graphical Model / Joint Inference
10
Joint Inference over evidence in a table

• Probabilistic Graphical Models

11
A graphical model for tables
Class
C2
C3
C1
Team
Chicago
Philadelphia
Houston
San Antonio
R21
R31
R11
R12
R22
R32
R13
R23
R33
Instance
12
Parameterized graphical model
Captures interaction between row values
 
 
 
R33
R11
R12
R13
R21
R22
R23
R31
R32
 
 
 
Row value
Factor Node
C2
C1
C3
Function that captures the affinity between the
column headers and row values
Variable Node Column header
Captures interaction between column headers
 
13
Nice, but Will it work ?
14
Evaluation

• Dataset of gt 6000 tables 2
• Compare our accuracy against our baseline system
  and the results in 2
• Use Mean Average Precision 3 to compare a
  ranked list of possible classes/entities
  against a ranked list obtained from human
  evaluators
• Experiment with datasets from www.data.gov

15
References

1. Cafarella, M. J., Halevy, A., Wang, D. Z., Wu,
   E., Zhang, Y., 2008. Webtables exploring the
   power of tables on the web. Proc. VLDB Endow.1
   (1), 538-549.
2. Limaye, G., Sarawagi, S., Chakrabarti, S.
   Annotating and searching web tables using
   entities, types and relationships. In Proc. 36th
   Int. Conf. on Very Large Databases (2010)
3. Salton, G., Mcgill, M.J. Introduction to Modern
   Information Retrieval. McGraw-Hill, Inc., New
   York (1986)

16

• Thank You !
• Questions ?
• varish1_at_cs.umbc.edu_at_varish
• Webhttp//ebiq.org/h/Varish/Mulwad