[developers] help with interpreting redwoods.mem
oe at ifi.uio.no
Mon Jul 25 23:59:34 CEST 2011
those features are indeed all generated by template 1, where
derivations are ‘featurized’ all the way down to their leafs—
hence one sees surface forms in some of the features.
On 25. juli 2011, at 23.48, Spencer T Rarrick <rarricks at uw.edu> wrote:
> I am beginning some work with parse disambiguation using the model
> in "redwoods.mem", and I have a question about how to interpret the
> meaning of features. At the top In features.lisp (http://svn.emmtee.net/trunk/lingo/lkb/src/tsdb/lisp/features.lisp
> ), the comments list several feature templates numbered 1-4, 10, 11,
> etc. However all of the features found in redwoods.mem show a "1"
> after the open bracket "[", indicating that they should be from
> feature template 1. This seems strange as many (but not all) of the
> features include a string at the end (such as "well"), even though
> the description of template 1 does not mention lexicalization (while
> template 3 does).
> If anyone is familiar with the model, could you please let me know
> if my interpretation of the file is correct? If not, how can one
> distinguish between the various feature templates in the file?
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the developers