BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Sensing well-being using heterogeneous smartphone data and stance 
 identification in social media conversations - Maria Liakata (Warwick Univ
 ersity)
DTSTART:20161124T110000Z
DTEND:20161124T120000Z
UID:TALK68126@talks.cam.ac.uk
CONTACT:Mohammad Taher Pilehvar
DESCRIPTION:In the first part of my talk I will describe a new problem of 
 predicting affect and well-being scales in a real-world setting of heterog
 eneous\, longitudinal and non-synchronous textual as well as non-linguisti
 c data that can be harvested from on-line media and mobile phones. We desc
 ribe the method for collecting the heterogeneous longitudinal data\, how f
 eatures are extracted to address missing information and differences in te
 mporal alignment\, and how the latter are combined using multi-kernel lear
 ning to yield promising predictions of affect and well-being. \n\nIn the s
 econd part of my talk I will discuss rumour stance classification as a seq
 uential task. Rumour stance classification\, the task that determines if e
 ach tweet in a collection discussing a rumour is supporting\, denying\, qu
 estioning or simply commenting on the rumour\, has been attracting substan
 tial interest. We introduce a novel approach that makes use of the sequenc
 e of transitions observed in tree-structured conversation threads in Twitt
 er. The conversation threads are formed by harvesting users' replies to on
 e another\, which results in a nested tree-like structure. Previous work a
 ddressing the stance classification task has treated each tweet as a separ
 ate unit. Here we analyse tweets by virtue of their position in a sequence
  and test two sequential classifiers\, Linear-Chain CRF and Tree CRF\, eac
 h of which makes different assumptions about the conversational structure.
  We experiment with eight Twitter datasets\, collected during breaking new
 s\, and show that exploiting the sequential structure of Twitter conversat
 ions achieves significant improvements over the non-sequential methods. 
LOCATION:GR05\, English Faculty\, 9 West Road (Sidgwick Site)
END:VEVENT
END:VCALENDAR