[Corpora-List] Chinese and English POS

[Linas Vepstas]

2009/11/3 Geoffrey Sampson <grs2 at sussex.ac.uk>:

>  What one can do is to try to _impose_ rules that succeed in
> determining a unique PoS tagging in as many debatable cases as possible

--Well, humour and poetry both sometimes rely on ambiguity.
"Your money or your life! Take my wife, please!"

Is there a bibliography on how to best discover and represent
ambiguity (in POS, and also in syntax)?  I think this is interesting
for both practical purposes,and for theoretical enjoyment.

>>From the semantic-stack point of view, its useful to generate multiple
parses for ambiguous sentences, and then pass these up to
other software layers, which can perhaps apply reasoning, logic,
or various word-sense disambiguation algos to help disambiguate
between parses.  A useful engineering question is how to compactly
represent the ambiguity, so as not to chew up disk space and
processing time with oodles of similar parses.

>>From a more theoretical view, having some automated ambiguity
score might be interesting for understanding humor, but also as
a measure for misleadingly ambiguous legal contracts, or even
unintentional ambiguity in witness testimony or in scientific
papers.

--linas

_______________________________________________
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora