BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:A Standard Document Score for Information Retrieval  - Ronan Cummi
 ns\, University of Greenwich
DTSTART:20131129T120000Z
DTEND:20131129T130000Z
UID:TALK48926@talks.cam.ac.uk
CONTACT:Tamara Polajnar
DESCRIPTION:\nRanking functions are a crucial component in information ret
 rieval systems. These functions assign a score to each document in a colle
 ction with respect to a static query. Regularising these document scores a
 cross different queries is a difficult challenge that has many potential a
 pplications.\n\nIn this talk I will outline a number of different retrieva
 l tasks for which score regularisation is important. I propose and outline
  a standard document retrieval score based on term-frequencies. I will sho
 w that the standardisation approach adopted automatically creates a measur
 e of term-specifity. An analysis shows that this measure is highly correla
 ted with the traditional idf measure\, and furthermore\, suggests a novel 
 interpretation of idf-like measures.\n\nFinally\, I will present an evalua
 tion on a number of different datasets that shows that the standard docume
 nt score is comparable with the BM25 retrieval function in terms of effect
 iveness. However\, an advantage of the standard document score is that und
 er certain conditions\, the document scores output from the model are comp
 arable across different queries and collections.
LOCATION:FW26\, Computer Laboratory
END:VEVENT
END:VCALENDAR