Experiments with Corpus-based LFG Specialization
Nicola Cancedda, Christer Samuelsson
Sophisticated grammar formalisms, such as LFG, allow concisely capturing complex linguistic phenomena.
The powerful operators provided by such formalisms can however introduce spurious ambiguity, making parsing
inefficient. A simple form of corpus-based grammar pruning is evaluated experimentally on two wide-coverage
grammars, one English and one French. Speedups of up to a factor 6 were obtained, at a cost in grammatical
coverage of about 13%. A two-stage architecture allows achieving significant speedups without introducing
additional parse failures.
Proc. of the Sixth Applied Natural Language Processing Conference (ANLP 2000), Seattle, WA, 2000
anlp00.pdf (125.38 kB)
anlp00.ps (136.65 kB)