But what do you mean by a domain? From what I've seen, it could be as broad as 'sport' or as narrow as 'WSJ stock price reports' To be scientific about domains, we need to talk scientifically about differences between them. The only candidate method that I know of for doing that represents the domain by a corpus of texts from it. So a prior question to 'how do we adapt for a new domain' is 'how do we compare corpora' - which, oddly, is a question that has not received much attention.<br>
<br>Adam Kilgarriff<br><br><div class="gmail_quote">On 20 May 2010 15:53, JIANG Jing <span dir="ltr"><<a href="mailto:jingjiang@smu.edu.sg">jingjiang@smu.edu.sg</a>></span> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
[ please distribute - apologies for multiple postings ]<br>
<br>
========================================================================<br>
=====<br>
Final Call for Papers<br>
<br>
ACM Transactions on Intelligent Systems and Technology (ACM TIST)<br>
<br>
=== Special Issue on Domain Adaptation in Natural Language Processing<br>
===<br>
<br>
<a href="http://tist.acm.org/" target="_blank">http://tist.acm.org/</a><br>
<br>
Full Paper Submission Deadline: June 1, 2010<br>
Review Notification: September 1, 2010<br>
Final Manuscript: November 1, 2010<br>
Publication Date: December 2010<br>
========================================================================<br>
=====<br>
<br>
------------------<br>
Topics of Interest<br>
------------------<br>
<br>
Over the past two decades, supervised learning methods have been<br>
successfully applied to many natural language processing problems such<br>
as syntactic parsing, information extraction and machine translation.<br>
However, a major drawback of supervised learning methods is their heavy<br>
reliance on the quality and size of annotated training corpora, which<br>
are highly labor-intensive to create. It is well understood that when<br>
test data comes from a different domain and thus has a different<br>
distribution than the training data, performance of learning-based<br>
systems can drop substantially. In natural language processing, this<br>
domain adaptation problem has been reported for various tasks including<br>
word sense disambiguation, parsing, named entity recognition and<br>
sentiment analysis, to name just a few. Although this is a fundamental<br>
problem with statistical learning, it only started gaining much<br>
attention in recent years.<br>
<br>
The objective of this special issue is to provide a venue to highlight<br>
some of the recent advances in developing domain adaptive techniques for<br>
natural language processing and related areas such as information<br>
retrieval and text mining, with an emphasis on applications and systems.<br>
Topics of interest include but are not limited to<br>
<br>
* novel domain adaptation techniques and applications designed<br>
with a focus on NLP problems<br>
* evaluation of general domain adaptation systems applied to<br>
specific NLP problems<br>
* adaptation of NLP tools to handle noisy text data such as email<br>
and blogs<br>
* cross-lingual adaptation techniques and systems<br>
* analysis and comparison between domain adaptation and other<br>
related problems such as semi-supervised learning and active learning<br>
for NLP problems<br>
* techniques and systems for measuring domain relatedness and<br>
learning from multiple domains in NLP<br>
* domain adaptive NLP techniques applied to multi-disciplinary<br>
domains such as medicine and bioinformatics areas<br>
<br>
-----------<br>
Submissions<br>
-----------<br>
<br>
On-Line Submission (will be available before June 1, 2010):<br>
<a href="http://mc.manuscriptcentral.com/tist" target="_blank">http://mc.manuscriptcentral.com/tist</a> (please select "Special Issue:<br>
Domain Adaptation in Natural Language Processing" as the manuscript<br>
type)<br>
<br>
Details of the journal and manuscript preparation are available on the<br>
website:<br>
<a href="http://tist.acm.org/" target="_blank">http://tist.acm.org/</a><br>
<br>
Each paper will be peer-reviewed by at least three reviewers.<br>
<br>
---------------<br>
Important Dates<br>
---------------<br>
<br>
Full Paper Submission Deadline: June 1, 2010<br>
Review Notification: September 1, 2010<br>
Final Manuscript: November 1, 2010<br>
Publication Date: December 2010<br>
<br>
-------------<br>
Guest Editors<br>
-------------<br>
<br>
Hal Daume III (University of Utah)<br>
Jing Jiang (Singapore Management University), Special Issue Contact<br>
(jingjiang at smu dot edu dot sg)<br>
Sinno Jialin Pan (Hong Kong University of Science and Technology)<br>
Masashi Sugiyama (Tokyo Institute of Technology)<br>
<br>
<br>
_______________________________________________<br>
Corpora mailing list<br>
<a href="mailto:Corpora@uib.no">Corpora@uib.no</a><br>
<a href="http://mailman.uib.no/listinfo/corpora" target="_blank">http://mailman.uib.no/listinfo/corpora</a><br>
</blockquote></div><br><br clear="all"><br>-- <br>================================================<br>Adam Kilgarriff <a href="http://www.kilgarriff.co.uk">http://www.kilgarriff.co.uk</a> <br>
Lexical Computing Ltd <a href="http://www.sketchengine.co.uk">http://www.sketchengine.co.uk</a><br>Lexicography MasterClass Ltd <a href="http://www.lexmasterclass.com">http://www.lexmasterclass.com</a><br>
Universities of Leeds and Sussex <a href="mailto:adam@lexmasterclass.com">adam@lexmasterclass.com</a><br>================================================<br>