BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Automatically Creating Reading Lists with Topical PageRank - James
  Jardine\, University of Cambridge
DTSTART:20120224T120000Z
DTEND:20120224T130000Z
UID:TALK36624@talks.cam.ac.uk
CONTACT:Ekaterina Kochmar
DESCRIPTION:We present an algorithm for creating reading lists - lists of 
 papers given\nby an expert to a novice\, designed to bring the novice up t
 o speed in a\ncertain area. Our algorithm uses a variant of PageRank that 
 is age-corrected\nand sensitive to the mixture of papers' topics as determ
 ined by the LDA\ntopic model.\nWhen compared to a gold standard of reading
  lists which we collected from\nexperts\, our algorithm outperforms three 
 currently used keyword-based search\nengines: Lucene\, Google Scholar and 
 the Google-indexed ACL Anthology. As\nevaluation metrics we use F-measure\
 , as well as a new evaluation metric\nspecific to reading lists which we i
 ntroduce here. It estimates the degree\nof substitutability of expert pape
 rs by system-found ones by the number of\nlinks in the citation network be
 tween them.  We also evaluate on the task of\nreference list reintroductio
 n. When reintroducing the reference list of\nthousands of papers\, our uns
 upervised algorithm performs on a par with the\ncurrent state-of-the-art m
 ethod\, which is supervised.
LOCATION:FW26\, Computer Laboratory
END:VEVENT
END:VCALENDAR