Faster Parsing by Supertagger Adaptation

Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics, pages 345–355, Uppsala, Sweden, 11-16 July 2010. c⃝2010 Association for Computational Linguistics Faster Parsing by Supertagger Adaptation Jonathan K. Kummerfelda Jessika Roesnerb Tim Dawborna James Haggertya James R. Currana∗ Stephen Clarkc∗ School of Information Technologiesa Department of Computer Scienceb Computer Laboratoryc University of Sydney University of Texas at Austin University of Cambridge NSW 2006, Australia Austin, TX, USA Cambridge CB3 0FD, UK james@it.usyd.edu.aua∗ stephen.clark@cl.cam.ac.ukc∗ Abstract We propose a novel self-training method for a parser which uses a lexicalised gram- mar and supertagger, focusing on increas- ing the speed of the parser rather than its accuracy. The idea is to train the su- pertagger on large amounts of parser out- put, so that the supertagger can learn to supply the supertags that the parser will eventually choose as part of the highest- scoring derivation. Since the supertag- ger supplies fewer supertags overall, the parsing speed is increased. We demon- strate the effectiveness of the method us- ing a CCG supertagger and parser, obtain- ing significant speed increases on newspa- per text with no loss in accuracy. We also show that the method can be used to adapt the CCG parser to new domains, obtain- ing accuracy and speed improvements for Wikipedia and biomedical text. 1 Introduction In many NLP tasks and applications, e.g. distribu- tional similarity (Curran, 2004) and question an- swering (Dumais et al., 2002), large volumes of text and detailed syntactic information are both critical for high performance. To avoid a trade- off between these two, we need to increase parsing speed, but without losing accuracy. Parsing with lexicalised grammar formalisms, such as Lexicalised Tree Adjoining Grammar and Combinatory Categorial Grammar (CCG; Steed- man, 2000), can be made more efficient using a supertagger. Bangalore and Joshi (1999) call su- pertagging almost parsing because of the signifi- cant reduction in ambiguity which occurs once the supertags have been assigned. In this paper, we focus on the CCG parser and supertagger described in Clark and Curran (2007). Since the CCG lexical category set used by the su- pertagger is much larger than the Penn Treebank POS tag set, the accuracy of supertagging is much lower than POS tagging; hence the CCG supertag- ger assigns multiple supertags1 to a word, when the local context does not provide enough infor- mation to decide on the correct supertag. The supertagger feeds lexical categories to the parser, and the two interact, sometimes using mul- tiple passes over a sentence. If a spanning analy- sis cannot be found by the parser, the number of lexical categories supplied by the supertagger is increased. The supertagger-parser interaction in- fluences speed in two ways: first, the larger the lexical ambiguity, the more derivations the parser must consider; second, each further pass is as costly as parsing a whole extra sentence. Our goal is to increase parsing speed without loss of accuracy. The technique we use is a form of self-training, in which the output of the parser …

Content truncated to 500 words