![]() |
|
Part of the purpose of this website is to make it easy for people to experiment with RST. Various sorts of tools can help. Tools for analysts This website benefited from one of these, programmed by Mick O'Donnell. The tool can help a great deal in constructing diagrams of RST analyses. An earlier version of it was used to produce the diagrams on this website. Since 1999 it has been extensively improved and debugged. Over 100 changes have been made, and it is now more ready than ever for widespread use. We gratefully acknowledge Mick O'Donnel's work that makes this possible. Other annotation tools Mick O’Donnell's tool was modified by Daniel Marcu, and is available from his web site: http://www.isi.edu/licensed-sw/RSTTool/ LaTeX tools David Reitter has created a tool to generate RST-style diagrams using the LaTeX text processing software. The package produces an RST tree and marks its corresponding text with the appropriate span labels: http://www.reitter-it-media.de/compling/rst/. Tables for accessing RST relation definitions Part of this website. For use in manual analysis of texts. Relation Lists and Definitions. Analysis files There are RST analysis files for the texts and diagrams exhibited on the website. There are 15 text analyses currently on the web site, with the Mother Teresa text having 3 analyses, so there are 17 analyses in all. The collection can be downloaded, currently for Windows only, as a self-expanding EXE file. The file can be used only with Mick O'Donnell's RSTTool (see above).
To download the file, use this link: RST Analyses. The file is tiny, less than 50 K. This file is also available under Analyses, together with individual pdf and gif files of each of the 17 texts. RST Corpus The RST Corpus is a collection of Wall Street Journal articles annotated using (a version of) RST by Lynn Carlson, Daniel Marcu and Mary Ellen Okurowski. The corpus is available through the Linguistic Data Consortium, free for members, and at a cost for non-members (Catalog number: LDC2002T07). Further information: Discourse Relations Reference Corpus The materials in the Discourse Relations Reference Corpus are taken from three different sources: texts from this web site; annotated Wall Street Journal articles from the RST Discourse Treebank (see above); and review texts from the SFU Review Corpus. The documents in each of the subcorpora have been annotated with RSTTool (see above). Although the background to all subcorpora is Rhetorical Structure Theory, and they have been annotated with RSTTool, we believe that the corpus is useful to anyone interested in discourse relations, from whatever perspective. The annotations provide rich information on what relations are more common; how they are commonly signalled; and how relations are distributed in different genres. A description and the full collection are available from the Discourse Relations Reference Corpus page. Text generation research In addition, there are various kinds of computer programs that have been oriented by RST. Several efforts have benefited from the orientation of RST, some following the first published work much more closely than others. To learn about text generation and find a demonstration of text generation on line on the net, go here: Text Generation.
A set of slides that can serve as a basic introduction to RST. The slides come out of courses taught by Manfred Stede and Maite Taboada. You are free to use and/or modify them. We would appreciate an acknowledgement of the source.
|
||||
©2005-2009 William C. Mann, Maite Taboada. All rights reserved. |