News and info about Machine Translation at Microsofthttp://blogs.msdn.com/christw/atom.xmlCommunity Server2007-02-18T23:30:00Zhttp://blogs.msdn.com/christw/archive/2009/09/05/optimizing-a-statistical-mt-system-using-shared-data.aspx2009-09-05T17:28:00Z2009-09-05T17:28:00ZA very substantial feature of statistical machine translation (SMT) systems is the ability to train the system on different bodies of text, and this way customizing it and optimizing it for a certain style and vocabulary. In MT land we call documents of related style and terminology a "domain". The most valuable training data for an SMT system is what we call "parallel data": the same document in two languages. The format that makes parallel data immediately useful is a translation memory (TM) file...(<a href="http://blogs.msdn.com/christw/archive/2009/09/05/optimizing-a-statistical-mt-system-using-shared-data.aspx">read more</a>)<img src="http://blogs.msdn.com/aggbug.aspx?PostID=9891792" width="1" height="1">christwhttp://blogs.msdn.com/members/christw.aspxhttp://blogs.msdn.com/christw/archive/2009/09/05/notes-from-the-mt-summit-in-ottawa.aspx2009-09-05T16:41:00Z2009-09-05T16:41:00ZLast week I attended the MT Summit in Ottawa ( http://summitxii.amtaweb.org/),mostly in the commercial user track. Compared to the previous two AMTA conferences I attended, it was striking to see the attention on MT-aided translation and translator productivity, in all 3 tracks at the conference (research, commercial users, and government users). What was also striking is that nobody has actual solid data on post-editing productivity in statistically significant numbers. Memorable papers related...(<a href="http://blogs.msdn.com/christw/archive/2009/09/05/notes-from-the-mt-summit-in-ottawa.aspx">read more</a>)<img src="http://blogs.msdn.com/aggbug.aspx?PostID=9891777" width="1" height="1">christwhttp://blogs.msdn.com/members/christw.aspxhttp://blogs.msdn.com/christw/archive/2007/02/18/microsoft-knowledge-base.aspx2007-02-19T10:54:00Z2007-02-19T10:54:00Z<P>My team is responsible for the MT engine that creates the foreign language translations of knowledge base (KB) articles on <A href="http://support.microsoft.com/" mce_href="http://support.microsoft.com/">http://support.microsoft.com</A>. Each KB article has at&nbsp;the bottom a feeedback field allowing the reader to rate the article. I pay close attention to this feedback to determine the usefulness of the MT engine for a task like understanding a technical document.</P> <P>Feel free to let me know right here of any significant issues or recurring problems you see with the KB translation.</P><img src="http://blogs.msdn.com/aggbug.aspx?PostID=1712817" width="1" height="1">christwhttp://blogs.msdn.com/members/christw.aspxhttp://blogs.msdn.com/christw/archive/2007/02/18/citation-in-the-french-press.aspx2007-02-19T10:30:00Z2007-02-19T10:30:00Z<P class=MsoPlainText style="MARGIN: 0in 0in 0pt">I am cited in last weekend's issue of the French newspaper “Libération”: <A href="http://www.liberation.fr/transversales/weekend/234277.FR.php" mce_href="http://www.liberation.fr/transversales/weekend/234277.FR.php">http://www.liberation.fr/transversales/weekend/234277.FR.php</A> under the title "Babelweb". Pretty positive about MT in general. </P> <P class=MsoPlainText style="MARGIN: 0in 0in 0pt" mce_keep="true">&nbsp;</P> <P class=MsoPlainText style="MARGIN: 0in 0in 0pt">Thierry Fontenelle blogged about the article in his post&nbsp; <SPAN style="FONT-SIZE: 10pt; COLOR: navy; FONT-FAMILY: 'Arial','sans-serif'; mso-fareast-font-family: Calibri; mso-fareast-theme-font: minor-latin; mso-ansi-language: EN-US; mso-fareast-language: EN-US; mso-bidi-language: AR-SA"><A href="http://blogs.msdn.com/correcteurorthographiqueoffice/">http://blogs.msdn.com/correcteurorthographiqueoffice/</A>. </SPAN></P><img src="http://blogs.msdn.com/aggbug.aspx?PostID=1712708" width="1" height="1">christwhttp://blogs.msdn.com/members/christw.aspx