RST-Style Discourse Parsing and Its Applications in Discourse Analysis. Feng, V. W. Ph.D. Thesis, Department of Computer Science, University of Toronto, 2015. abstract bibtex Discourse parsing is the task of identifying the relatedness and the particular discourse relations among various discourse units in a text. In particular, among various theoretical frameworks of discourse parsing, I am interested in Rhetorical Structure Theory (RST). I hypothesize that, given its ultimate success, discourse parsing can provide a general solution for use in many downstream applications.
This thesis is composed of two major parts. First, I overview my work on discourse segmentation and discourse tree-building, which are the two primary components of RST-style discourse parsing. Evaluated on the RST Discourse Treebank (RST-DT), both of my discourse segmenter and tree-builder achieve the state-of-the-art performance.
Later, I discuss the application of discourse relations to some specific tasks in the analysis of discourse, including the evaluation of coherence, the identification of authorship, and the detection of deception. In particular, I propose to use a set of application-neutral features, which are derived from the discourse relations extracted by my discourse parser, and compare the performance of these application-neutral features against the classic application-specific approaches to each of these tasks. On the first two tasks, experimental results show that discourse relation features by themselves often perform as well as those classic application-specific features, and the combination of these two kinds of features usually yields further improvement. These results provide strong evidence for my hypothesis that discourse parsing is able to provide a general solution for the analysis of discourse. However, we failed to observe a similar effectiveness of discourse parsing on the third task, the detection of deception. I postulate that this might be due to several confounding factors of the task itself.
@phdthesis{FengPhD,
author = "Vanessa Wei Feng",
title = "RST-Style Discourse Parsing and Its Applications in Discourse Analysis",
school = "Department of Computer Science, University of Toronto",
year = "2015",
abstract = "Discourse parsing is the task of identifying the
relatedness and the particular discourse relations
among various discourse units in a text. In
particular, among various theoretical frameworks of
discourse parsing, I am interested in Rhetorical
Structure Theory (RST). I hypothesize that, given
its ultimate success, discourse parsing can provide
a general solution for use in many downstream
applications. <P> This thesis is composed of two
major parts. First, I overview my work on discourse
segmentation and discourse tree-building, which are
the two primary components of RST-style discourse
parsing. Evaluated on the RST Discourse Treebank
(RST-DT), both of my discourse segmenter and
tree-builder achieve the state-of-the-art
performance. <P> Later, I discuss the application
of discourse relations to some specific tasks in the
analysis of discourse, including the evaluation of
coherence, the identification of authorship, and the
detection of deception. In particular, I propose to
use a set of application-neutral features, which are
derived from the discourse relations extracted by my
discourse parser, and compare the performance of
these application-neutral features against the
classic application-specific approaches to each of
these tasks. On the first two tasks, experimental
results show that discourse relation features by
themselves often perform as well as those classic
application-specific features, and the combination
of these two kinds of features usually yields
further improvement. These results provide strong
evidence for my hypothesis that discourse parsing is
able to provide a general solution for the analysis
of discourse. However, we failed to observe a
similar effectiveness of discourse parsing on the
third task, the detection of deception. I postulate
that this might be due to several confounding
factors of the task itself.",
download = "http://ftp.cs.toronto.edu/pub/gh/Feng-thesis-2015.pdf"
}
Downloads: 0
{"_id":"NoqB5YtMqtb67MbMq","bibbaseid":"feng-rststylediscourseparsinganditsapplicationsindiscourseanalysis-2015","downloads":0,"creationDate":"2016-10-19T19:03:53.835Z","title":"RST-Style Discourse Parsing and Its Applications in Discourse Analysis","author_short":["Feng, V. W."],"year":2015,"bibtype":"phdthesis","biburl":"http://www.cs.toronto.edu/compling/all_bib.bib","bibdata":{"bibtype":"phdthesis","type":"phdthesis","author":[{"firstnames":["Vanessa","Wei"],"propositions":[],"lastnames":["Feng"],"suffixes":[]}],"title":"RST-Style Discourse Parsing and Its Applications in Discourse Analysis","school":"Department of Computer Science, University of Toronto","year":"2015","abstract":"Discourse parsing is the task of identifying the relatedness and the particular discourse relations among various discourse units in a text. In particular, among various theoretical frameworks of discourse parsing, I am interested in Rhetorical Structure Theory (RST). I hypothesize that, given its ultimate success, discourse parsing can provide a general solution for use in many downstream applications. <P> This thesis is composed of two major parts. First, I overview my work on discourse segmentation and discourse tree-building, which are the two primary components of RST-style discourse parsing. Evaluated on the RST Discourse Treebank (RST-DT), both of my discourse segmenter and tree-builder achieve the state-of-the-art performance. <P> Later, I discuss the application of discourse relations to some specific tasks in the analysis of discourse, including the evaluation of coherence, the identification of authorship, and the detection of deception. In particular, I propose to use a set of application-neutral features, which are derived from the discourse relations extracted by my discourse parser, and compare the performance of these application-neutral features against the classic application-specific approaches to each of these tasks. On the first two tasks, experimental results show that discourse relation features by themselves often perform as well as those classic application-specific features, and the combination of these two kinds of features usually yields further improvement. These results provide strong evidence for my hypothesis that discourse parsing is able to provide a general solution for the analysis of discourse. However, we failed to observe a similar effectiveness of discourse parsing on the third task, the detection of deception. I postulate that this might be due to several confounding factors of the task itself.","download":"http://ftp.cs.toronto.edu/pub/gh/Feng-thesis-2015.pdf","bibtex":"@phdthesis{FengPhD,\n author = \"Vanessa Wei Feng\",\n title = \"RST-Style Discourse Parsing and Its Applications in Discourse Analysis\",\n school = \"Department of Computer Science, University of Toronto\",\n year = \"2015\",\n abstract = \"Discourse parsing is the task of identifying the\n relatedness and the particular discourse relations\n among various discourse units in a text. In\n particular, among various theoretical frameworks of\n discourse parsing, I am interested in Rhetorical\n Structure Theory (RST). I hypothesize that, given\n its ultimate success, discourse parsing can provide\n a general solution for use in many downstream\n applications. <P> This thesis is composed of two\n major parts. First, I overview my work on discourse\n segmentation and discourse tree-building, which are\n the two primary components of RST-style discourse\n parsing. Evaluated on the RST Discourse Treebank\n (RST-DT), both of my discourse segmenter and\n tree-builder achieve the state-of-the-art\n performance. <P> Later, I discuss the application\n of discourse relations to some specific tasks in the\n analysis of discourse, including the evaluation of\n coherence, the identification of authorship, and the\n detection of deception. In particular, I propose to\n use a set of application-neutral features, which are\n derived from the discourse relations extracted by my\n discourse parser, and compare the performance of\n these application-neutral features against the\n classic application-specific approaches to each of\n these tasks. On the first two tasks, experimental\n results show that discourse relation features by\n themselves often perform as well as those classic\n application-specific features, and the combination\n of these two kinds of features usually yields\n further improvement. These results provide strong\n evidence for my hypothesis that discourse parsing is\n able to provide a general solution for the analysis\n of discourse. However, we failed to observe a\n similar effectiveness of discourse parsing on the\n third task, the detection of deception. I postulate\n that this might be due to several confounding\n factors of the task itself.\", \n download = \"http://ftp.cs.toronto.edu/pub/gh/Feng-thesis-2015.pdf\"\n}\n\n\n","author_short":["Feng, V. W."],"key":"FengPhD","id":"FengPhD","bibbaseid":"feng-rststylediscourseparsinganditsapplicationsindiscourseanalysis-2015","role":"author","urls":{},"downloads":0},"search_terms":["rst","style","discourse","parsing","applications","discourse","analysis","feng"],"keywords":[],"authorIDs":[],"dataSources":["2vBSdbWEoTEQZtb6g"]}