Entity Extraction for Query Interpretation Patrick Pantel ǂ Query Representation and Understanding SIGIR July 23, 2010 Collaborators: Alpa Jain, Ana-Maria Popescu, Arkady Borkovsky, Eric Crestan, Hadar Shemtov, Marco Pennacchiotti, Nicolas Torzec, Vishnu Vyas ǂ Now at Microsoft Research
23
Embed
Entity Extraction for Query Interpretation Patrick Pantel ǂ
Entity Extraction for Query Interpretation Patrick Pantel ǂ. Query Representation and Understanding SIGIR July 23, 2010. Collaborators : Alpa Jain, Ana-Maria Popescu, Arkady Borkovsky , Eric Crestan, Hadar Shemtov, Marco Pennacchiotti, Nicolas Torzec, Vishnu Vyas - PowerPoint PPT Presentation
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Entity Extractionfor Query Interpretation
Patrick Pantelǂ
Query Representation and Understanding SIGIR
July 23, 2010
Collaborators:Alpa Jain, Ana-Maria Popescu, Arkady Borkovsky, Eric Crestan, Hadar Shemtov, Marco Pennacchiotti, Nicolas Torzec, Vishnu Vyas
extractor [Pantel et al., EMNLP 2009]• Given a small set of seeds for a
given class, find distributionally similar candidate instances
Distributional Extractor
Nicole KidmanAl PacinoTom Hanks
Web
anna gunnolivier gueriteetomas von bromssenharry jonesjudy mathesonrobert keithmariah o'brienstarring dennis quaidnoah beery jrfederico castelluccioadienne shellybetty morangeorge takaijo anne worleyruth hampton
rex hagonalex fonggene burkemiguel hermoso arnaoeiko andocharles mccaughanyukijiro hotarualec christiedame wendy hillerjohn waynearthur lakesir herbert beerbohm treetonya wrightlori saunders
- 10 -Entity Extraction July 2010
S1
SK
S2
KE
nK
E2
KE
1
FG1 FG2 FGm
KB
FEATURE GENERATORS
RANKER
KN
OW
LED
GE
EX
TRA
CTO
RS
AG
GR
EG
ATO
R
MODELER
DECODER
Feature Generators
- 11 -Entity Extraction July 2010
Feature sets• 4 feature families• 5 feature types• 402 features
Web600M pages web crawl Query log1 year of queries (top 1M)Web TableFrom 600M pages web crawlWikipedia2M articles 2008 dump