Shallow processing

[Bouma G.]

Katia,

> 2. I would also be interested in any work concerning the automatic (e.g. corpora-based) implementation of an HPSG lexicon. I know of most subcategorization papers (Brent, Manning, Carroll, Korhonen, Roland/Jurafsky, de Lima, Buchholz etc).
>

For a wide-coverage HPSG-style Dutch grammar we used two lexical resources,
a version of the Celex lexical database providing valency information and the
Dutch Parole lexicon, to extract subcategorization and dependency information.
An overview of the number of verbal `dependency frames' extracted and an estimation
of coverage can be found in

Gosse Bouma, Extracting dependency frames from existing lexical resources,

www.let.rug.nl/alfa/papers/papers/papers.html

There is also a paper there describing how we use dependency relations for
tree-banking and evaluation:

Gosse Bouma, Gertjan van Noord, and Robert Malouf,
Alpino: Wide-coverage computational analysis of Dutch.

best,

Gosse.

--
Gosse Bouma, Alfa-informatica, RUG, Postbus 716, 9700 AS Groningen
gosse at let.rug.nl      tel. +31-50-3635937      fax  +31-50-3636855