Efficient HPSG parsing with supertagging and CFG-filtering

  • Authors:
  • Takuya Matsuzaki;Yusuke Miyao;Jun'ichi Tsujii

  • Affiliations:
  • Department of Computer Science, University of Tokyo, Bunkyo-ku, Tokyo, Japan;Department of Computer Science, University of Tokyo, Bunkyo-ku, Tokyo, Japan;Department of Computer Science, University of Tokyo, Bunkyo-ku, Tokyo, Japan and School of Computer Science, University of Manchester and National Center for Text Mining

  • Venue:
  • IJCAI'07 Proceedings of the 20th international joint conference on Artifical intelligence
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

An efficient parsing technique for HPSG is presented. Recent research has shown that supertagging is a key technology to improve both the speed and accuracy of lexicalized grammar parsing. We show that further speed-up is possible by eliminating non-parsable lexical entry sequences from the output of the supertagger. The parsability of the lexical entry sequences is tested by a technique called CFG-filtering, where a CFG that approximates the HPSG is used to test it. Those lexical entry sequences that passed through the CFG-filter are combined into parse trees by using a simple shift-reduce parsing algorithm, in which structural ambiguities are resolved using a classifier and all the syntactic constraints represented in the original grammar are checked. Experimental results show that our system gives comparable accuracy with a speed-up by a factor of six (30 msec/sentence) compared with the best published result using the same grammar.