Abstract
This paper describes Grammar Learning by Partition Search, a general method for automatically constructing grammars for a range of parsing tasks. Given a base grammar, a training corpus, and a parsing task, Partition Search constructs an optimised probabilistic context-free grammar by searching a space of nonterminal set partitions, looking for a partition that maximises parsing performance and minimises grammar size. The method can be used to optimise grammars in terms of size and performance, or to adapt existing grammars to new parsing tasks and new domains. This paper reports an example application to optimising a base grammar extracted from the Wall Street Journal Corpus. Partition Search improves parsing performance by up to 5.29%, and reduces grammar size by up to 16.89%. Parsing results are better than in existing treebank grammar research, and compared to other grammar compression methods, Partition Search has the advantage of achieving compression without loss of grammar coverage.
Original language | English |
---|---|
Title of host publication | Proceedings of the LREC 2002 workshop on event modelling for multilingual document linking |
Place of Publication | Amsterdam/Philadelphia |
Publisher | John Benjamins Publishing Company |
Pages | 0-0 |
Number of pages | 1 |
Publication status | Published - 1 Jan 2002 |
Event | Proceedings of the LREC 2002 workshop on event modelling for multilingual document linking - Las Palmas, Spain Duration: 1 Jan 2002 → … |
Workshop
Workshop | Proceedings of the LREC 2002 workshop on event modelling for multilingual document linking |
---|---|
Period | 1/01/02 → … |
Keywords
- Natural language generation
- Partition searching