Recent Changes - Search:

Publications

Résumé

 Class projects

 Links

 Archive


> Robin Stewart






Login

Thesis 06 Abstract

Main.Thesis06Abstract History

Hide minor edits - Show changes to output

October 30, 2008, at 03:23 PM by 198.151.161.130 -
Changed lines 9-11 from:
||border=0 width=90% cellpadding=15
|| ||A collection of recorded and transcribed telephone conversations clearly demonstrates the universality of small talk and other socially-motivated utterances.  Building on theories about the linguistics of conversational speech, I consider various ways of describing each utterance, including which words were used, their
part-of-speech, and the proximity to the beginning of the conversation.  In order to better understand which of these features are most useful, I create a system for automatically distinguishing between on- and off-topic utterances and compare its performance when using different combinations of these features.  The central hypothesis is that conversational speech contains sufficient low-level clues to separate on- and off-topic utterances with an automatic classifier.  I find that the overall structure of conversations is predictable, and automatic classification can indeed be done with better-than-chance accuracy.  But distinguishing more reliably between on- and off-topic utterances will probably require deeper knowledge of the context and overall topic. ||
to:
->A collection of recorded and transcribed telephone conversations clearly demonstrates the universality of small talk and other socially-motivated utterances.  Building on theories about the linguistics of conversational speech, I consider various ways of describing each utterance, including which words were used, their part-of-speech, and the proximity to the beginning of the conversation.  In order to better understand which of these features are most useful, I create a system for automatically distinguishing between on- and off-topic utterances and compare its performance when using different combinations of these features.  The central hypothesis is that conversational speech contains sufficient low-level clues to separate on- and off-topic utterances with an automatic classifier.  I find that the overall structure of conversations is predictable, and automatic classification can indeed be done with better-than-chance accuracy.  But distinguishing more reliably between on- and off-topic utterances will probably require deeper knowledge of the context and overall topic.
May 23, 2008, at 01:01 PM by 128.30.44.143 -
Changed line 3 from:
'''[[http://www.robinstewart.com/research/papers/stewart06thesis.pdf | Automatic Identification of Off-Topic Regions of Conversation]]'''\\
to:
!!!'''Automatic Identification of Off-Topic Regions of Conversation'''
Changed lines 8-9 from:
!! Abstract:
to:
Abstract:
Added lines 12-13:
'''[[http://www.robinstewart.com/research/papers/stewart06thesis.pdf | Download Thesis]]''' (pdf)
July 11, 2006, at 09:54 PM by 207.178.20.6 -
Changed line 3 from:
'''Automatic Identification of Off-Topic Regions of Conversation'''\\
to:
'''[[http://www.robinstewart.com/research/papers/stewart06thesis.pdf | Automatic Identification of Off-Topic Regions of Conversation]]'''\\
July 11, 2006, at 09:53 PM by 207.178.20.6 -
Changed lines 10-23 from:
->A collection of recorded and transcribed telephone conversations clearly demonstrates
->the universality of small talk and other socially-motivated utterances. Building on
->theories about the linguistics of conversational speech, I consider various ways of de-
->scribing
each utterance, including which words were used, their part-of-speech, and
->the proximity to the beginning of the conversation. In order to better understand
->which of these features are most useful, I create a system for automatically distin-
->guishing between on- and off-topic utterances and compare its performance when
->using different combinations of these
features. The central hypothesis is that conver-
->sational
speech contains sufficient low-level clues to separate on- and off-topic utter-
->ances
with an automatic classifier. I find that the overall structure of conversations is
->predictable, and automatic classification can indeed be done with better-than-chance
->accuracy. But distinguishing more reliably between on- and off-topic utterances will
->probably require deeper knowledge of the context and overall topic.
to:
||border=0 width=90% cellpadding=15
|| ||A collection of recorded and transcribed telephone conversations clearly
demonstrates the universality of small talk and other socially-motivated utterances.  Building on theories about the linguistics of conversational speech, I consider various ways of describing each utterance, including which words were used, their part-of-speech, and the proximity to the beginning of the conversation.  In order to better understand which of these features are most useful, I create a system for automatically distinguishing between on- and off-topic utterances and compare its performance when using different combinations of these features.  The central hypothesis is that conversational speech contains sufficient low-level clues to separate on- and off-topic utterances with an automatic classifier.  I find that the overall structure of conversations is predictable, and automatic classification can indeed be done with better-than-chance accuracy.  But distinguishing more reliably between on- and off-topic utterances will probably require deeper knowledge of the context and overall topic. ||
July 11, 2006, at 09:50 PM by 207.178.20.6 -
Changed lines 10-23 from:
A collection of recorded and transcribed telephone conversations clearly demonstrates
the universality of small talk and other socially-motivated utterances. Building on
theories about the linguistics of conversational speech, I consider various ways of de-
scribing each utterance, including which words were used, their part-of-speech, and
the proximity to the beginning of the conversation. In order to better understand
which of these features are most useful, I create a system for automatically distin-
guishing between on- and off-topic utterances and compare its performance when
using different combinations of these features. The central hypothesis is that conver-
sational speech contains sufficient low-level clues to separate on- and off-topic utter-
ances with an automatic classifier. I find that the overall structure of conversations is
predictable, and automatic classification can indeed be done with better-than-chance
accuracy. But distinguishing more reliably between on- and off-topic utterances will
probably require deeper knowledge of the context and overall topic.
to:
->A collection of recorded and transcribed telephone conversations clearly demonstrates
->the universality of small talk and other socially-motivated utterances. Building on
->theories about the linguistics of conversational speech, I consider various ways of de-
->scribing each utterance, including which words were used, their part-of-speech, and
->the proximity to the beginning of the conversation. In order to better understand
->which of these features are most useful, I create a system for automatically distin-
->guishing between on- and off-topic utterances and compare its performance when
->using different combinations of these features. The central hypothesis is that conver-
->sational speech contains sufficient low-level clues to separate on- and off-topic utter-
->ances with an automatic classifier. I find that the overall structure of conversations is
->predictable, and automatic classification can indeed be done with better-than-chance
->accuracy. But distinguishing more reliably between on- and off-topic utterances will
->probably require deeper knowledge of the context and overall topic.
July 11, 2006, at 09:42 PM by 207.178.20.6 -
Changed lines 1-7 from:
(:title Abstract :)

!! Automatic Identification of Off-Topic Regions of Conversation\\
May 22, 2006\\
Senior Thesis\\
Williams College
to:
(:notitle:)

'''Automatic Identification of Off-Topic Regions of Conversation'''\\
''Senior Thesis\\
Williams College\\
May 22, 2006''

!! Abstract:

Changed lines 22-24 from:
probably require deeper knowledge of the context and overall topic.
to:
probably require deeper knowledge of the context and overall topic.

[[<<]]
July 11, 2006, at 09:39 PM by 207.178.20.6 -
Added lines 1-20:
(:title Abstract :)

!! Automatic Identification of Off-Topic Regions of Conversation\\
May 22, 2006\\
Senior Thesis\\
Williams College

A collection of recorded and transcribed telephone conversations clearly demonstrates
the universality of small talk and other socially-motivated utterances. Building on
theories about the linguistics of conversational speech, I consider various ways of de-
scribing each utterance, including which words were used, their part-of-speech, and
the proximity to the beginning of the conversation. In order to better understand
which of these features are most useful, I create a system for automatically distin-
guishing between on- and off-topic utterances and compare its performance when
using different combinations of these features. The central hypothesis is that conver-
sational speech contains sufficient low-level clues to separate on- and off-topic utter-
ances with an automatic classifier. I find that the overall structure of conversations is
predictable, and automatic classification can indeed be done with better-than-chance
accuracy. But distinguishing more reliably between on- and off-topic utterances will
probably require deeper knowledge of the context and overall topic.
Edit - History - Print - Recent Changes - Search
Page last modified on October 30, 2008, at 03:23 PM