lecture18-learning-ranking-handouts-6-per

Butonlyrecentlyhavemlknowledgeavailabilityof

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: (successful)
 Informa)on
Retrieval
   Ranking
Model:
f(ψi) [Yunbo
Cao,
Jun
Xu,
Tie‐Yan
Liu,
Hang
Li,
Yalou
Huang,
 Hsiao‐Wuen
Hon
SIGIR
2006]
   A
Ranking
SVM
model
already
works
well
 [Herbrich
et
al.
1999,
2000;
Joachims
et
al.
2002]
   Using
things
like
vector
space
model
scores
as
features

   As
we
shall
see,
it
outperforms
them
in
evalua)ons
   But
it
does
not
model
important
aspects
of
prac)cal
 IR
well

   This
paper
addresses
two
customiza)ons
of
the
 Ranking
SVM
to
fit
an
IR
u)lity
model
 5 5/30/11 Introduc)on to Informa)on Retrieval Introduc)on to Informa)on Retrieval The
ranking
SVM
fails
to
model
the
IR
problem
 well…
 1.  Correctly
ordering
the
most
relevant
documents
is
 crucial
to
the
success
of
an
IR
system,
while
 misordering
less
relevant
results
magers
ligle
   The
ranking
SVM
considers
all
ordering
viola)ons
as
the
 same
 2.  Some
queries
have
many
(somewhat)
relevant
 document...
View Full Document

This document was uploaded on 02/26/2014.

Ask a homework question - tutors are online