5 March: DCLRS Fri March 7th, 2008, 16:00, Prof. Walter Daelemans -- DCU
Dublin Computational Linguistics Research Seminar: Index of March 2008 | Dublin Computational Linguistics Research Seminar - Index of year: 2008 | Full index
Dublin Computational Linguistics Research Seminar Series DCLRS-2007/8
Fri March 7th, 2008, 16:00, L2.21, School of Computing, DCU
Prof. Walter Daelemans (University of Antwerp)
TITLE:
Text Analysis and Machine Learning for Personality and Author
Detection from Text
ABSTRACT:
I will present work in progress on author and personality detection in
the Personae corpus, a corpus of essays produced by 145 different
authors with associated personality profiles (assigned using the
Meyers Briggs Type Indicator). I wil discuss how the combination of
text analysis (using memory-based shallow parsing) and machine
learning can to some extent allow the identification of personality
from text, and identification of the textual indicators most
predictive for this task. On the basis of author identification
experiments on the same corpus, I will argue that previous work on
authorship attribution may have been overoptimistic, and that features
working well for distinguishing between a small set of authors may not
work for discriminating between large sets of authors.
All welcome!
-----------------------------------------------
DCLRS is a joint seminar series run by DCU, DIT, TCD and UCD since
1996/7. DCLRS 2007/8 is hosted by the National Centre for Language
Technology NCLT, School of Computing, DCU.
Dublin Computational Linguistics Research Seminar - Index of March 2008 | Index of year: 2008 | Full index