Rhetorical structure theory

From HandWiki
Short description: Theory of text organization
detail of a Tree of Knowledge after Diderot & d'Alembert's Encyclopédie, by Chrétien Frédéric Guillaume Roth
Information mapping
Topics and fields
Node–link approaches
See also

Rhetorical structure theory (RST) is a theory of text organization that describes relations that hold between parts of text. It was originally developed by William Mann, Sandra Thompson, Christian M. I. M. Matthiessen and others at the University of Southern California's Information Sciences Institute (ISI) and defined in a 1988 paper.[1][2][3] The theory was developed as part of studies of computer-based text generation. Natural language processing researchers later began using RST in automatic summarization and other applications. It explains coherence by postulating a hierarchical, connected structure of texts,[3] which are labeled using a small, predefined inventory of relation types - for example, one part of a text may provide an elaboration on another part, provide background or specify a cause for another.[1]

In the 2000s, following the release of the first large-scale dataset implementing the theory, the RST Discourse Treebank (RST-DT),[4] Daniel Marcu demonstrated the feasibility of practical applications of RST to discourse parsing and summarization at ISI.[5] Originally limited to written text, subsequent work in the 2010s expanded RST to spoken language analysis,[6] and the framework has been applied to a variety of languages including Farsi,[7] German,[8] Mandarin Chinese,[9] Russian[10] and Spanish.[11] Following the introduction of Transformers, LLMs have been applied to automatic RST parsing,[12] with results approaching human performance on parsing text in English.

Rhetorical relations

Rhetorical relations, also called coherence or discourse relations, are paratactic (coordinate) or hypotactic (subordinate) relations that hold across two or more text spans.[13] The logical arrangement of relations in a text contributes to its coherence by connecting different propositions in a relational structure. RST using rhetorical relations provides a systematic way for an analyst to analyze the underlying intention of a text. The analysis is usually built by reading the text and constructing a tree using the relations. The following example is a title and summary, appearing at the top of an article in Scientific American magazine (adapted from Ramachandran and Anstis, 1986). The original text, broken into numbered units, is:[3]

File:RST example.png
Diagram of an RST analysis
  1. [Title:] The Perception of Apparent Motion
  2. [Abstract:] When the motion of an intermittently seen object is ambiguous
  3. the visual system resolves confusion
  4. by applying some tricks
  5. that reflect a built-in knowledge of properties of the physical world.


In the figure, the numbers 1-5 show the corresponding units from the text above. Unit 5 provides an "elaboration" on unit 4, and therefore constitutes a less prominent satellite of unit 4, which acts as a nucleus for the relation. Units 4-5 form a relation "Means", explaining the means by which the visual system resolves confusion. Unit 3 is the Central Discourse Unit (CDU) of the text, since all units point to it directly or indirectly. Similarly units 1 and 2 form "preparation" and "circumstance" relations relative to their nuclei. Groups of units which serve as a satellite or nucleus together are called complex discourse units, and always span a set of adjacent EDUs.

Nuclearity in discourse

RST establishes two different types of units. Nuclei are considered as the most important parts of text whereas satellites contribute to the nuclei and are secondary. Nucleus contains basic information and satellite contains additional information about nucleus. The satellite is often incomprehensible without nucleus, whereas a text where satellites have been deleted can be understood to a certain extent.

Hierarchy in the analysis

RST relations are applied recursively in a text, until all units in that text are constituents in an RST relation. The result of such analyses is that RST structure are typically represented as trees, with one top level relation that encompasses other relations at lower levels.

Why RST?

  1. From linguistic point of view, RST proposes a different view of text organization than most linguistic theories.
  2. RST points to a tight relation between relations and coherence in text
  3. From a computational point of view, it provides a characterization of text relations that has been implemented in different systems and for applications as text generation[14] and summarization.[15]

In design rationale

Computer scientists Ana Cristina Bicharra Garcia and Clarisse Sieckenius de Souz have used RST as the basis of a design rationale system called ADD+.[16][17] In ADD+, RST is used as the basis for the rhetorical organization of a knowledge base, in a way comparable to other knowledge representation systems such as issue-based information system (IBIS).[17] Similarly, RST has been used in representation schemes for argumentation.[18][19][20]

See also

References

  1. ↑ 1.0 1.1 Mann, William C.; Thompson, Sandra A. (1988). "Rhetorical structure theory: toward a functional theory of text organization". Text: Interdisciplinary Journal for the Study of Discourse 8 (3): 243–281. doi:10.1515/text.1.1988.8.3.243. http://www.cis.upenn.edu/~nenkova/Courses/cis700-2/rst.pdf. Retrieved 1 November 2017. 
  2. ↑ Matthiessen, Christian M. I. M. (June 2005). "Remembering Bill Mann". Computational Linguistics 31 (2): 161–171. doi:10.1162/0891201054224002. http://www.aclweb.org/anthology/J05-2001. Retrieved 1 November 2017. 
  3. ↑ 3.0 3.1 3.2 Taboada, Maite; Mann, William C. (June 2006). "Rhetorical structure theory: looking back and moving ahead". Discourse Studies 8 (3): 423–459. doi:10.1177/1461445606061881. https://www.sfu.ca/~mtaboada/docs/publications/Taboada_Mann_RST_Part1.pdf. 
  4. ↑ Carlson, Lynn; Marcu, Daniel; Okurowski, Mary Ellen (2003). "Building a discourse-tagged corpus in the framework of rhetorical structure theory". in Kuppevelt, Jan van; Smith, Ronnie W.. Current and new directions in discourse and dialogue. Text, speech, and language technology. 22. Dordrecht; Boston: Kluwer Academic Publishers. pp. 85–112. doi:10.1007/978-94-010-0019-2_5. ISBN 978-1402016141. OCLC 53097055. https://www.isi.edu/~marcu/papers/sigdialbook2002.pdf. 
  5. ↑ Marcu, Daniel (2000). The theory and practice of discourse parsing and summarization. Cambridge, Mass.: MIT Press. ISBN 978-0262133722. OCLC 43811223. https://www.isi.edu/~marcu/papers/discourse-book/. 
  6. ↑ Zeldes, Amir (2017-09-01). "The GUM corpus: creating multilayer resources in the classroom" (in en). Language Resources and Evaluation 51 (3): 581–612. doi:10.1007/s10579-016-9343-x. ISSN 1574-0218. https://doi.org/10.1007/s10579-016-9343-x. 
  7. ↑ "dblp: Persian Rhetorical Structure Theory." (in en). https://dblp.uni-trier.de/rec/journals/corr/abs-2106-13833.html. 
  8. ↑ Stede, Manfred; Neumann, Arne (2014-05). Calzolari, Nicoletta; Choukri, Khalid; Declerck, Thierry et al.. eds. "Potsdam Commentary Corpus 2.0: Annotation for Discourse Research". Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC'14) (Reykjavik, Iceland: European Language Resources Association (ELRA)): 925–929. https://aclanthology.org/L14-1468/. 
  9. ↑ Peng, Siyao; Liu, Yang Janet; Zeldes, Amir (2022-11). He, Yulan; Ji, Heng; Li, Sujian et al.. eds. "GCDT: A Chinese RST Treebank for Multigenre and Multilingual Discourse Parsing". Proceedings of the 2nd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 12th International Joint Conference on Natural Language Processing (Volume 2: Short Papers) (Online only: Association for Computational Linguistics): 382–391. doi:10.18653/v1/2022.aacl-short.47. https://aclanthology.org/2022.aacl-short.47/. 
  10. ↑ Toldova, Svetlana; Pisarevskaya, Dina; Ananyeva, Margarita; Kobozeva, Maria; Nasedkin, Alexander; Nikiforova, Sofia; Pavlova, Irina; Shelepov, Alexey (2017-09). Taboada, M.; da Cunha, I.; Maziero, E.G. et al.. eds. "Rhetorical relations markers in Russian RST Treebank". Proceedings of the 6th Workshop on Recent Advances in RST and Related Formalisms (Santiago de Compostela, Spain: Association for Computational Linguistics): 29–33. doi:10.18653/v1/W17-3604. ISBN 978-1-945626-78-4. https://aclanthology.org/W17-3604/. 
  11. ↑ da Cunha, Iria; Torres-Moreno, Juan-Manuel; Sierra, Gerardo; Cabrera-Diego, Luis-Adrián; Castro-Rolón, Brenda-Gabriela; Rolland Bartilotti, Juan-Miguel (2011-09). Mitkov, Ruslan; Angelova, Galia. eds. "The RST Spanish Treebank On-line Interface". Proceedings of the International Conference Recent Advances in Natural Language Processing 2011 (Hissar, Bulgaria: Association for Computational Linguistics): 698–703. https://aclanthology.org/R11-1101/. 
  12. ↑ Maekawa, Aru; Hirao, Tsutomu; Kamigaito, Hidetaka; Okumura, Manabu (2024-03). Graham, Yvette; Purver, Matthew. eds. "Can we obtain significant success in RST discourse parsing by using Large Language Models?". Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers) (St. Julian's, Malta: Association for Computational Linguistics): 2803–2815. doi:10.18653/v1/2024.eacl-long.171. https://aclanthology.org/2024.eacl-long.171/. 
  13. ↑ Taboada, Maite (2009). "Implicit and explicit coherence relations". in Renkema, Jan. Discourse, of course: an overview of research in discourse studies. Amsterdam; Philadelphia: John Benjamins Publishing Company. pp. 127–140. doi:10.1075/z.148.13tab. ISBN 9789027232588. OCLC 276996573. https://www.sfu.ca/~mtaboada/docs/publications/Taboada_Implicit_Explicit.pdf. 
  14. ↑ "RST and text generation". http://ccl.pku.edu.cn/doubtfire/NLP/Text_Generation/RST/RST%20and%20Text%20Generation.htm. 
  15. ↑ Uzêda, Vinícius Rodrigues; Pardo, Thiago Alexandre Salgueiro; Nunes, Maria das Graças Volpe (November 2008). "Evaluation of automatic text summarization methods based on rhetorical structure theory". Eighth International Conference on Intelligent Systems Design and Applications: Kaohsiung, Taiwan, 26–28 November 2008. ISDA'08. 2. Piscataway, NJ: IEEE. pp. 389–394. doi:10.1109/ISDA.2008.289. ISBN 978-0-7695-3382-7. http://www.icmc.usp.br/pessoas/taspardo/ISDA2008-UzedaEtAl.pdf. Retrieved 1 November 2017. 
  16. ↑ Garcia, Ana Cristina Bicharra; Souz, Clarisse Sieckenius de (April 1997). "ADD+: Including rhetorical structures in active documents". Artificial Intelligence for Engineering Design, Analysis and Manufacturing 11 (2): 109–124. doi:10.1017/S0890060400001906. https://www.cambridge.org/core/services/aop-cambridge-core/content/view/329DA7DE0F67BABE209E8DF82E7D4901/S0890060400001906a.pdf/add-including-rhetorical-structures-in-active-documents.pdf. 
  17. ↑ 17.0 17.1 Regli, William C.; Hu, Xiaochun; Atwood, Michael; Sun, Wei (December 2000). "A survey of design rationale systems: approaches, representation, capture and retrieval". Engineering with Computers 16 (3–4): 209–235. doi:10.1007/PL00013715. http://gicl.cs.drexel.edu/images/d/d4/Eng-w-Computers-Design-Rationale-Dec-2000.pdf. 
  18. ↑ Green, Nancy L. (August 2009). "Representation of argumentation in text with rhetorical structure theory". Argumentation 24 (2): 181–196. doi:10.1007/s10503-009-9169-4. 
  19. ↑ Green, Nancy L. (November 2015). "Annotating evidence-based argumentation in biomedical text". 2015 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), Washington, DC, USA, 9–12 November 2015. Piscataway, NJ: IEEE. pp. 922–929. doi:10.1109/BIBM.2015.7359807. ISBN 978-1-4673-6799-8. OCLC 972619754. 
  20. ↑ Mitrović, Jelena; O'Reilly, Cliff; Mladenović, Miljana; Handschuh, Siegfried (January 2017). "Ontological representations of rhetorical figures for argument mining". Argument & Computation 8 (3): 267–287. doi:10.3233/AAC-170027. 



Retrieved from "https://handwiki.org/wiki/index.php?title=Rhetorical_structure_theory&oldid=4298113"