國立虎尾科技大學 |

Cross-Paragraph Discourse Structure in Rhetorical Structure Theory Parsing and Treebanking for Chinese and English.

Record Type:	Language materials, manuscript : Monograph/item
Title/Author:	Cross-Paragraph Discourse Structure in Rhetorical Structure Theory Parsing and Treebanking for Chinese and English./
Author:	Peng, Siyao.
Description:	1 online resource (168 pages)
Notes:	Source: Dissertations Abstracts International, Volume: 84-11, Section: B.
Contained By:	Dissertations Abstracts International84-11B.
Subject:	Linguistics. -
Online resource:	click for full text (PQDT)
ISBN:	9798379454388

Cross-Paragraph Discourse Structure in Rhetorical Structure Theory Parsing and Treebanking for Chinese and English.
Peng, Siyao.

Cross-Paragraph Discourse Structure in Rhetorical Structure Theory Parsing and Treebanking for Chinese and English. - 1 online resource (168 pages)

Source: Dissertations Abstracts International, Volume: 84-11, Section: B.

Thesis (Ph.D.)--Georgetown University, 2023.

Includes bibliographical references

Hierarchical discourse structures benefit Natural Language Understanding tasks, such as text summarization and sentiment analysis. Rhetorical Structure Theory (RST) is particularly significant at the macro levels, such as between paragraphs. Moreover, RST parsing at the macro level is more challenging than at the micro level, where intra-sentential relations are the easiest to identify.Despite a dozen RST datasets available in multiple languages, a sizeable Chinese RST corpus still needs to be created. Moreover, there awaits an in-depth analysis regarding how much RST associates with macro-level structures and how much parsing performance deteriorates at the macro level and across genres. Using English and Chinese as examples, this dissertation examines how macro-level discourse relations are presented in RST and whether state-of-the-art RST parsers capture them properly. Firstly, I create the largest Chinese RST Corpus, namely Georgetown Chinese Discourse Treebank (GCDT), an open-source treebank with 50 medium-to-long documents from five different genres. I present basic statistics and highlight annotation decisions for Mandarin Chinese. I believe this sizeable multi-genre RST corpus can promote discourse analysis and RST parsing in Chinese and across languages.Secondly, I examine the association between paragraphs and RST trees from three aspects: a) studying how the lengths of EDU, sentence, and paragraph segments differ across genres and corpora; b) assessing whether or not paragraphs are fully contained in RST subtrees and 3) analyzing the distribution of intra- versus inter-paragraph relations across corpora and genres.Thirdly, I conduct parsing experiments on Chinese GCDT and English GUM using a state-of-the-art multilingual RST parser. I present both datasets' benchmark monolingual and multilingual parsing scores and boost the performance by pretraining and automatic translation. Moreover, I show that SOTA parsers are unsatisfactory in some genres and the inter-paragraph scenario.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2024

Mode of access: World Wide Web

ISBN: 9798379454388Subjects--Topical Terms:

557829
Linguistics.
Subjects--Index Terms:

ChineseIndex Terms--Genre/Form:

554714
Electronic books.

Cross-Paragraph Discourse Structure in Rhetorical Structure Theory Parsing and Treebanking for Chinese and English.
LDR:03555ntm a22004337 4500 001 1142207
005 20240416125327.5
006 m o d
007 cr mn ---uuuuu
008 250605s2023 xx obm 000 0 eng d
020 $a 9798379454388
035 $a (MiAaPQ)AAI30419618
035 $a AAI30419618
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Peng, Siyao. $3 1466475
245 1 0 $a Cross-Paragraph Discourse Structure in Rhetorical Structure Theory Parsing and Treebanking for Chinese and English.
264 0 $c 2023
300 $a 1 online resource (168 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 84-11, Section: B.
500 $a Advisor: Zeldes, Amir;Schneider, Nathan.
502 $a Thesis (Ph.D.)--Georgetown University, 2023.
504 $a Includes bibliographical references
520 $a Hierarchical discourse structures benefit Natural Language Understanding tasks, such as text summarization and sentiment analysis. Rhetorical Structure Theory (RST) is particularly significant at the macro levels, such as between paragraphs. Moreover, RST parsing at the macro level is more challenging than at the micro level, where intra-sentential relations are the easiest to identify.Despite a dozen RST datasets available in multiple languages, a sizeable Chinese RST corpus still needs to be created. Moreover, there awaits an in-depth analysis regarding how much RST associates with macro-level structures and how much parsing performance deteriorates at the macro level and across genres. Using English and Chinese as examples, this dissertation examines how macro-level discourse relations are presented in RST and whether state-of-the-art RST parsers capture them properly. Firstly, I create the largest Chinese RST Corpus, namely Georgetown Chinese Discourse Treebank (GCDT), an open-source treebank with 50 medium-to-long documents from five different genres. I present basic statistics and highlight annotation decisions for Mandarin Chinese. I believe this sizeable multi-genre RST corpus can promote discourse analysis and RST parsing in Chinese and across languages.Secondly, I examine the association between paragraphs and RST trees from three aspects: a) studying how the lengths of EDU, sentence, and paragraph segments differ across genres and corpora; b) assessing whether or not paragraphs are fully contained in RST subtrees and 3) analyzing the distribution of intra- versus inter-paragraph relations across corpora and genres.Thirdly, I conduct parsing experiments on Chinese GCDT and English GUM using a state-of-the-art multilingual RST parser. I present both datasets' benchmark monolingual and multilingual parsing scores and boost the performance by pretraining and automatic translation. Moreover, I show that SOTA parsers are unsatisfactory in some genres and the inter-paragraph scenario.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2024
538 $a Mode of access: World Wide Web
650 4 $a Linguistics. $3 557829
650 4 $a Computer science. $3 573171
650 4 $a Rhetoric. $3 567738
650 4 $a Bilingual education. $3 1148431
653 $a Chinese
653 $a Corpus linguistics
653 $a Discourse parsing
653 $a English
653 $a Multilingual
653 $a Rhetorical Structure Theory
653 $a Georgetown Chinese Discourse Treebank
655 7 $a Electronic books. $2 local $3 554714
690 $a 0290
690 $a 0984
690 $a 0282
690 $a 0681
710 2 $a ProQuest Information and Learning Co. $3 1178819
710 2 $a Georgetown University. $b Linguistics. $3 1182048
773 0 $t Dissertations Abstracts International $g 84-11B.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30419618 $z click for full text (PQDT)