Abstract

This paper describes and evaluates log-linear parsing models for Combinatory Categorial Grammar (CCG). A parallel implementation of the L-BFGS optimisation algorithm is described, which runs on a Beowulf cluster allowing the complete Penn Treebank to be used for estimation. We also develop a new efficient parsing algorithm for CCG which maximises expected recall of dependencies. We compare models which use all CCG derivations, including nonstandard derivations, with normal-form models. The performances of the two models are comparable and the results are competitive with existing wide-coverage CCG parsers.

Citations

1243	Numerical Optimization - Nocedal, Wright - 2006
846	A Maximum Entropy Approach to Natural Language Processing - Berger, Pietra, et al. - 1996
396	New Statistical Parser Based on Bigram Lexical De-pendencies - Collins - 2006
396	The syntactic process - Steedman - 2000
336	F: Shallow parsing with conditional random fields - Sha, Pereira
181	R: A Gaussian prior for smoothing maximum entropy models - SF, Rosenfeld - 1999
78	Parsing algorithms and metrics - Goodman - 1996
59	Maximum entropy estimation for feature forests - Miyao, Tsujii - 2002
56	2004. The Importance of Supertagging for Wide-Coverage CCG Parsing - Clark, Curran
38	Investigating gis and smoothing for maximum entropy taggers - Curran, Clark - 2003
36	Efficient normal-form parsing for combinatory categorial grammar - Eisner - 1996
24	Log-linear models for wide-coverage CCG parsing - Clark, Curran - 2003
6	Parsing with generative models of predicate-argument structure - 2003b
4	Data and Models for Statistical Parsing with Combinatory Categorial Grammar - 2003a