Revision: 16662
http://sourceforge.net/p/gate/code/16662
Author: adamfunk
Date: 2013-04-26 15:53:50 +0000 (Fri, 26 Apr 2013)
Log Message:
-----------
Documenting the recent addition of L- and U-tags to the conll format.
Modified Paths:
--------------
userguide/trunk/corpora.tex
Modified: userguide/trunk/corpora.tex
===================================================================
--- userguide/trunk/corpora.tex 2013-04-26 01:22:13 UTC (rev 16661)
+++ userguide/trunk/corpora.tex 2013-04-26 15:53:50 UTC (rev 16662)
@@ -1366,7 +1366,7 @@
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsect[sec:corpora:conll]{CoNLL/IOB Documents}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-GATE can read files of text annotated in the traditionall CoNLL or ``IOB''
+GATE can read files of text annotated in the traditional CoNLL or BIO/BILOU
format, typically used to represent POS tags and chunks and best known for
Conference on Natural Language
Learning\footnote{\url{http://ifarm.nl/signll/conll/}} tasks. The following
@@ -1399,9 +1399,11 @@
(beginning with \texttt{B-} and followed by zero or more matching \texttt{I-}
tags) produces an annotation whose type is determined by the rest of the tag
(\texttt{NP} or \texttt{VP} in the above example, but any string with no
-whitespace is acceptable), with a \emph{kind = chunk} feature. Other tags
+whitespace is acceptable), with a \emph{kind = chunk} feature. A chunk tag
+beginning with \texttt{L-} (\emph{last}) terminates the chunk, and a
\texttt{U-}
+(\emph{unigram}) tag produces a chunk annotation over one token. Other tags
produce annotations with the tag name as the type and a \emph{kind = token}
-feature.
+feature.
Every annotation derived from a tag has a \emph{column} feature whose
\texttt{int} value indicates the source column in the data (numbered from 0 for
This was sent by the SourceForge.net collaborative development platform, the
world's largest Open Source development site.
------------------------------------------------------------------------------
Try New Relic Now & We'll Send You this Cool Shirt
New Relic is the only SaaS-based application performance monitoring service
that delivers powerful full stack analytics. Optimize and monitor your
browser, app, & servers with just a few lines of code. Try New Relic
and get this awesome Nerd Life shirt! http://p.sf.net/sfu/newrelic_d2d_apr
_______________________________________________
GATE-cvs mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/gate-cvs
