<html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40"><head><meta http-equiv=Content-Type content="text/html; charset=utf-8"><meta name=Generator content="Microsoft Word 15 (filtered medium)"><style><!--
/* Font Definitions */
@font-face
{font-family:"Cambria Math";
panose-1:2 4 5 3 5 4 6 3 2 4;}
@font-face
{font-family:Calibri;
panose-1:2 15 5 2 2 2 4 3 2 4;}
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0in;
margin-bottom:.0001pt;
font-size:11.0pt;
font-family:"Calibri",sans-serif;}
a:link, span.MsoHyperlink
{mso-style-priority:99;
color:blue;
text-decoration:underline;}
a:visited, span.MsoHyperlinkFollowed
{mso-style-priority:99;
color:purple;
text-decoration:underline;}
p.msonormal0, li.msonormal0, div.msonormal0
{mso-style-name:msonormal;
mso-margin-top-alt:auto;
margin-right:0in;
mso-margin-bottom-alt:auto;
margin-left:0in;
font-size:11.0pt;
font-family:"Calibri",sans-serif;}
span.EmailStyle19
{mso-style-type:personal-reply;
font-family:"Calibri",sans-serif;
color:#4472C4;}
.MsoChpDefault
{mso-style-type:export-only;
font-size:10.0pt;}
@page WordSection1
{size:8.5in 11.0in;
margin:1.0in 1.0in 1.0in 1.0in;}
div.WordSection1
{page:WordSection1;}
--></style><!--[if gte mso 9]><xml>
<o:shapedefaults v:ext="edit" spidmax="1026" />
</xml><![endif]--><!--[if gte mso 9]><xml>
<o:shapelayout v:ext="edit">
<o:idmap v:ext="edit" data="1" />
</o:shapelayout></xml><![endif]--></head><body lang=EN-US link=blue vlink=purple><div class=WordSection1><p class=MsoNormal><span style='color:#4472C4'>That’s funny! I’ve tried to prompt ChatGPT to do some RST analyses with a few examples, and it seems to me too that it’s seen some more information about RST during fine-tuning than just what might be gleaned from discussions about RST on the web. Then again, it has probably seen a lot of GitHub in the Pile, and you could find all sorts of RST-related data there.<o:p></o:p></span></p><p class=MsoNormal><span style='color:#4472C4'><o:p> </o:p></span></p><p class=MsoNormal><span style='color:#4472C4'>In fairness to ChatGPT, the RST Discourse Treebank does have a fine-grained “example” relation subtype in the coarse “elaboration” class, so it’s not totally off-base calling it an RST relation.<o:p></o:p></span></p><p class=MsoNormal><span style='color:#4472C4'><o:p> </o:p></span></p><p class=MsoNormal><span style='color:#4472C4'>Best,<o:p></o:p></span></p><p class=MsoNormal><span style='color:#4472C4'>Amir<o:p></o:p></span></p><p class=MsoNormal><span style='color:#4472C4'>------------<o:p></o:p></span></p><p class=MsoNormal><span style='color:#4472C4'>Dr. Amir Zeldes<o:p></o:p></span></p><p class=MsoNormal><span style='color:#4472C4'>Assoc. Prof. of Computational Linguistics<o:p></o:p></span></p><p class=MsoNormal><span style='color:#4472C4'>Department of Linguistics<o:p></o:p></span></p><p class=MsoNormal><span style='color:#4472C4'>Georgetown University<o:p></o:p></span></p><p class=MsoNormal><span style='color:#4472C4'>1437 37th St. NW<o:p></o:p></span></p><p class=MsoNormal><span style='color:#4472C4'>Washington, DC 20057<o:p></o:p></span></p><p class=MsoNormal><span style='color:#4472C4'><o:p> </o:p></span></p><p class=MsoNormal><span style='color:#4472C4'><a href="https://gucorpling.org/amir">https://gucorpling.org/amir</a> <o:p></o:p></span></p><p class=MsoNormal><span style='color:#4472C4'><o:p> </o:p></span></p><p class=MsoNormal><span style='color:#4472C4'><o:p> </o:p></span></p><div><div style='border:none;border-top:solid #E1E1E1 1.0pt;padding:3.0pt 0in 0in 0in'><p class=MsoNormal><b>From:</b> Rstlist <rstlist-bounces@listserv.linguistlist.org> <b>On Behalf Of </b>Matt Chung<br><b>Sent:</b> Tuesday, October 24, 2023 8:15 PM<br><b>To:</b> rstlist@listserv.linguistlist.org<br><b>Subject:</b> [Rstlist] Exemplification - listed as a relation according to ChatGPT<o:p></o:p></p></div></div><p class=MsoNormal><o:p> </o:p></p><div><div><p class=MsoNormal>I found it very interesting that ChatGPT listed “exemplification" as one of the <o:p></o:p></p></div><div><p class=MsoNormal>relations in RST and made me wonder how they trained their model regarding RST <o:p></o:p></p></div><div><p class=MsoNormal>...<o:p></o:p></p></div><div><p class=MsoNormal><o:p> </o:p></p></div><div><p class=MsoNormal>Just now, I asked ChatGPT to “describe all the relations in rhetorical <o:p></o:p></p></div><div><p class=MsoNormal>structure theory” and the results it spat out made me raise my eyebrows. Not <o:p></o:p></p></div><div><p class=MsoNormal>only did the number of relations (twenty three of them) stand out, but the <o:p></o:p></p></div><div><p class=MsoNormal>results also included “Exemplification”, which I neither recalled (from the <o:p></o:p></p></div><div><p class=MsoNormal>paper - Rhetorical Structure Theory: A theory of text organization) nor saw <o:p></o:p></p></div><div><p class=MsoNormal>listed on the RST website[1]. Therefore, I followed up in ChatGPT with, “For <o:p></o:p></p></div><div><p class=MsoNormal>your last message, provide me the source or reference for exemplification being <o:p></o:p></p></div><div><p class=MsoNormal>one of the primary relations in RST”, to which it responded with:<o:p></o:p></p></div><div><p class=MsoNormal><o:p> </o:p></p></div><div><p class=MsoNormal>"I apologize for the oversight. As of my last training cut-off in January 2022, <o:p></o:p></p></div><div><p class=MsoNormal>"Exemplification" is not one of the primary relations originally posited by <o:p></o:p></p></div><div><p class=MsoNormal>Mann and Thompson in their foundational work on Rhetorical Structure Theory <o:p></o:p></p></div><div><p class=MsoNormal>(RST). The relations I listed were meant to be an illustrative overview of RST, <o:p></o:p></p></div><div><p class=MsoNormal>but my inclusion of "Exemplification" was an error on my part.<o:p></o:p></p></div><div><p class=MsoNormal><o:p> </o:p></p></div><div><p class=MsoNormal>For the most accurate and authoritative information on RST, you would typically <o:p></o:p></p></div><div><p class=MsoNormal>refer to Mann and Thompson's seminal papers on the topic. Always refer to <o:p></o:p></p></div><div><p class=MsoNormal>primary literature or trusted secondary sources for definitive lists or <o:p></o:p></p></div><div><p class=MsoNormal>specific details.”<o:p></o:p></p></div></div><div><p class=MsoNormal><o:p> </o:p></p></div><div><p class=MsoNormal><o:p> </o:p></p></div><div><p class=MsoNormal>[1] - <a href="http://www.sfu.ca/rst/01intro/intro.html">http://www.sfu.ca/rst/01intro/intro.html</a><o:p></o:p></p></div><p class=MsoNormal><o:p> </o:p></p><div><div><div><div><div><p class=MsoNormal><span style='color:black'>Cheers,<br>Matt Chung<br><a href="https://tiny.crossbill.co/m/O5hkNb4Sg0o">Founder | Crossbill LLC</a><br>"Helping companies design & deliver robust software systems"<br><a href="https://tiny.crossbill.co/m/O5hk2zxmu68">LinkedIn</a> | <a href="https://tiny.crossbill.co/mblog/O5kQCX7C1pG">Blog</a> | <a href="https://tiny.crossbill.co/mdod/O5k0MSa7F4B">Digital Organization Dad</a><o:p></o:p></span></p></div></div></div></div></div><p class=MsoNormal><o:p> </o:p></p></div></body></html>