"just-in-time" sub-grammar extraction

[Ann Copestake]

Hi Vlado,

maybe you'd like to expand on this a bit?  The work on subgrammars that
I know of does not aim for complete consistency with the results of the
bigger grammar - in fact, generally the point is to cut down on the
number of readings.  It seems to me that if you have an algorithm that
shows part of a large grammar isn't applicable to a particular
sentence, then the obvious thing to do is to use that information to
cut down the search space, which I would call filtering, rather than
extraction of a sub-grammar.  Obviously, in this case, you are
generally only excluding part of the grammar for consideration for some
part of the processing.  If you are interested in that, have a look at
Kiefer et al, ACL 1999, and also the recent special issue of Natural
Language Engineering 6(1) where several papers mention different
filters and discuss their effectiveness.  Extracting a CFG backbone, as
in LiLFeS, can be thought of as a filtering technique if you then go on
to do a full parse for the derivations licensed by the CFG.  For what
it is worth, my guess is that there's lots of scope still for improving
filtering techniques.

But I think this is not what you mean so maybe you'd like to be more
specific about what you had in mind.

Ann