A Trellis-based algorithm for estimating the parameters of a hidden stochastic context-free grammar
HLT '91 Proceedings of the workshop on Speech and Natural Language
Statistical parsing of messages
HLT '90 Proceedings of the workshop on Speech and Natural Language
Augmenting a hidden Markov model for phrase-dependent word tagging
HLT '89 Proceedings of the workshop on Speech and Natural Language
Hi-index | 0.00 |
A new algorithm is presented for estimating the parameters of a stochastic context-free grammar (SCFG) from ordinary unparsed text. Unlike the Inside/Outside (I/O) algorithm which requires a grammar to be specified in Chomsky normal form, the new algorithm can estimate an arbitrary SCFG without any need for transformation. The algorithm has worst-case cubic complexity in the length of a sentence and the number of nonterminals in the grammar. Instead of the binary branching tree structure used by the I/O algorithm, the new algorithm makes use of a trellis structure for computation. The trellis is a generalization of that used by the Baum-Welch algorithm which is used for estimating hidden stochastic regular grammars. The paper describes the relationship between the trellis and the more typical parse tree representation.