BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:A Bottom-up Approach to Sentence Ordering for Multi-document Summa
 rization - Danushka Bollegala - University of Tokyo
DTSTART:20100122T120000Z
DTEND:20100122T130000Z
UID:TALK22812@talks.cam.ac.uk
CONTACT:Laura Rimell
DESCRIPTION:Ordering information is a difficult but important task for\nap
 plications generating natural-language texts such as multi-document\nsumma
 rization\,\nquestion answering\, and concept-to-text generation. In multi-
 document\nsummarization\,\ninformation is selected from a set of source do
 cuments.\nHowever\, improper ordering of information in a summary can conf
 use the\nreader and\ndeteriorate the readability of the summary. Therefore
 \, it is vital to\nproperly order the information\nin multi-document summa
 rization. We present a bottom-up\napproach to arrange sentences extracted 
 for multi-document\nsummarization. To capture the association and order of
  two\ntextual segments (e.g. sentences)\, we define four criteria: chronol
 ogy\,\ntopical-closeness\, precedence\, and succession.\nThese criteria ar
 e integrated into a criterion by a supervised learning\napproach.\nWe repe
 atedly concatenate two textual segments into one segment based\non the cri
 terion\, until we obtain the overall segment with all sentences\narranged.
  We evaluate the sentence orderings produced by the proposed\nmethod and\n
 numerous baselines using subjective gradings as well as automatic\nevaluat
 ion measures.\nWe introduce the average continuity\, an automatic evaluati
 on measure of\nsentence ordering in a summary\,\nand investigate its appro
 priateness for this task.
LOCATION:SW01\, Computer Laboratory
END:VEVENT
END:VCALENDAR