<html xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:st1="urn:schemas-microsoft-com:office:smarttags" xmlns="http://www.w3.org/TR/REC-html40">
<head>
<meta http-equiv=Content-Type content="text/html; charset=us-ascii">
<meta name=Generator content="Microsoft Word 11 (filtered medium)">
<o:SmartTagType namespaceuri="urn:schemas-microsoft-com:office:smarttags"
name="PlaceType"/>
<o:SmartTagType namespaceuri="urn:schemas-microsoft-com:office:smarttags"
name="PlaceName"/>
<o:SmartTagType namespaceuri="urn:schemas-microsoft-com:office:smarttags"
name="place" downloadurl="http://www.5iantlavalamp.com/"/>
<!--[if !mso]>
<style>
st1\:*{behavior:url(#default#ieooui) }
</style>
<![endif]-->
<style>
<!--
/* Font Definitions */
@font-face
{font-family:"MS Mincho";
panose-1:2 2 6 9 4 2 5 8 3 4;}
@font-face
{font-family:"\@MS Mincho";
panose-1:2 2 6 9 4 2 5 8 3 4;}
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0in;
margin-bottom:.0001pt;
font-size:12.0pt;
font-family:"Times New Roman";}
a:link, span.MsoHyperlink
{color:blue;
text-decoration:underline;}
a:visited, span.MsoHyperlinkFollowed
{color:purple;
text-decoration:underline;}
p.MsoPlainText, li.MsoPlainText, div.MsoPlainText
{margin:0in;
margin-bottom:.0001pt;
font-size:12.0pt;
font-family:"Courier New";}
p.Computer1, li.Computer1, div.Computer1
{margin:0in;
margin-bottom:.0001pt;
font-size:10.0pt;
font-family:"Courier New";}
@page Section1
{size:8.5in 11.0in;
margin:1.0in 32.35pt 1.0in 32.35pt;}
div.Section1
{page:Section1;}
-->
</style>
</head>
<body lang=EN-US link=blue vlink=purple>
<div class=Section1>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>Marco,<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>You wrote:<o:p></o:p></span></font></p>
<p class=MsoPlainText style='margin-left:.5in'><font size=3 face="Courier New"><span
style='font-size:12.0pt'>I fully agree with you and Katrin that the major
challenge for our model <o:p></o:p></span></font></p>
<p class=MsoPlainText style='margin-left:.5in'><font size=3 face="Courier New"><span
style='font-size:12.0pt'>and its alternatives is to find convincing ways to
evaluate whether it <o:p></o:p></span></font></p>
<p class=MsoPlainText style='margin-left:.5in'><font size=3 face="Courier New"><span
style='font-size:12.0pt'>learned what it purports to learn.<o:p></o:p></span></font></p>
<p class=MsoPlainText style='margin-left:.5in'><font size=3 face="Courier New"><span
style='font-size:12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText style='margin-left:.5in'><font size=3 face="Courier New"><span
style='font-size:12.0pt'>Best regards,<o:p></o:p></span></font></p>
<p class=MsoPlainText style='margin-left:.5in'><font size=3 face="Courier New"><span
style='font-size:12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText style='margin-left:.5in'><font size=3 face="Courier New"><span
style='font-size:12.0pt'>Marco<o:p></o:p></span></font></p>
<p class=MsoPlainText style='margin-left:.5in'><font size=3 face="Courier New"><span
style='font-size:12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>I can provide you with tools to make layered dictionaries of English keywords
if that helps you make vectors of the layered vocabularies. Would that be
useful for you?<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>-Rich<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'> <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>Sincerely,<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>Rich Cooper<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>EnglishLogicKernel.com<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>Rich AT EnglishLogicKernel DOT com<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>9 4 9 \ 5 2 5 - 5 7 1 2<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>-----Original Message-----<br>
From: corpora-bounces@uib.no [mailto:corpora-bounces@uib.no] On Behalf Of Marco
Baroni<br>
Sent: Friday, January 28, 2011 3:06 PM<br>
To: Yorick Wilks<br>
Cc: Corpora@uib.no; Krishnamurthy,Ramesh; Roberto Zamparelli<br>
Subject: Re: [Corpora-List] Moving Lexical Semantics from Alchemy to Science</span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>Dear Prof. Wilks,<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>I am one of the co-authors of the paper that Katrin kindly mentioned <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>(thanks, Katrin!).<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>Similar ideas are currently being explored by others, including
Emiliano <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>Guevara, Daoud Clarke and colleagues, and Edward Grefenstette and <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>colleagues.<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>We are using a mathematical tool from the mid 19th century (matrices)
in <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>order to apply intuitions from early seventies formal semantics <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>(Montague and others) to corpus-based semantic models that were <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>developed in the early nineties (LSA, HAL, ...), so we are not very
posh <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>-- we are a tad musty, if anything.<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>We represent adjectives as matrices because they are a simple way to <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>encode a function from and onto vectors.<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>We are trying to capture, in "distributional semantics", the
intuition <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>(expressed by Montague and many others) that adjectives are functions <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>that map nouns onto other nouns, where what the function does crucially
<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>depends on the input noun (so that "rubber" -- seen as an
adjective -- <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>is a function that can have a different effect when it maps
"ball" onto <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>"rubber ball" from the one it has when it maps
"duck" onto "rubber duck").<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>Since nouns, in many corpus-based approaches, are represented as
vectors <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>of co-occurrence counts with collocates (documents), we treat
adjectives <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>as matrices that encode linear functions from and onto such vectors.<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>I am (partially) aware of the literature on Pathfinder and other
earlier <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>literature on measuring word proximity, but it does not seem to me to <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>tackle the same challenge. We are using word/construction proximity to <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>evaluate our method, but the core of what the method does is building <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>larger constituents (adj+noun) from simpler ones (noun), which seems <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>like something different from what Pathfinder does (what little I know <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>of it).<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>I fully agree with you and Katrin that the major challenge for our
model <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>and its alternatives is to find convincing ways to evaluate whether it <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>learned what it purports to learn.<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>Best regards,<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>Marco<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>-- <o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>Marco Baroni<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>Center for Mind/Brain Sciences (CIMeC)<o:p></o:p></span></font></p>
<p class=MsoPlainText><st1:place w:st="on"><st1:PlaceType w:st="on"><font
size=3 face="Courier New"><span style='font-size:12.0pt'>University</span></font></st1:PlaceType>
of <st1:PlaceName w:st="on">Trento</st1:PlaceName></st1:place><o:p></o:p></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>http://clic.cimec.unitn.it/marco<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'><o:p> </o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>_______________________________________________<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>Corpora mailing list<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>Corpora@uib.no<o:p></o:p></span></font></p>
<p class=MsoPlainText><font size=3 face="Courier New"><span style='font-size:
12.0pt'>http://mailman.uib.no/listinfo/corpora<o:p></o:p></span></font></p>
</div>
</body>
</html>