Working with a lattice of lexical hypotheses and an (über)tagger, we seek to develop a filtering function that discards unlikely hypotheses. The formalisation of the lexical filtering process may be found here.

One such filter function maps PTB tags output from the TNT tagger onto LE Types. Mappings may be derived intuitively from inspection of a confusion matrix detailing the choices of TNT with respect to LE types.

An alternative approach is to find mappings based on the preferred outcomes of lexical filtering, (i.e. gains in parser efficiency versus losses in parser accuracy and coverage). These outcomes may be approximated by examining the relations between TNT precision, TNT recall and the ambiguity of LE types.

Frequency of LE types in JH0:

Plots of the TNT performance on the most frequent LE types

http://dl.dropbox.com/u/680530/WeSearch/Lexical%20Filtering/roc.png

A plot of the precision vs. lexical items filtered for each handled LE type:

http://dl.dropbox.com/u/680530/WeSearch/Lexical%20Filtering/filtering.png

Effective threshold ranges for LE Types:

• Driden, R. (2009), Using Lexical Statistics to Improve HPSG Parsing, PhD Thesis, Saarland University

• Ninomiya, T., Matsuzaki, T., Miyao, Y., Tsuruoka, Y and Tsujii, J. (2010). HPSG Parsing with a Supertagger. In Bunt, H., Merlo, P., and Nivre, J. (Eds.): Trends in Parsing Technology: Dependency Parsing, Domain Adaptation, and Deep Parsing. Springer, pp. 243-256.