In Data Oriented Parsing (DOP), an annotated language corpus is used as a virtual stochastic grammar. An input string is parsed by combining subtrees from the corpus. As a consequence, one parse tree can usually be generated by several derivations that involve different subtrees. This leads to a statistics where the probability of a parse is equal to the sum of the probabilities of all its derivations. In (Scha, 1990) an informal introduction to DOP is given, while (Bod, 1992) provides a formalization of the theory. In this paper we show that the maximum probability parse can be estimated in polynomial time by applying Monte Carlo techniques. The model was tested on a set of hand-parsed strings from the Air Travel Information System (ATIS) corpus. Preliminary experiments yield 96% test set parsing accuracy.
[1]
Mitchell P. Marcus.
Very Large Annotated Database of American English
,
1990,
HLT.
[2]
Philip Resnik,et al.
Probabilistic Tree-Adjoining Grammar as a Framework for Statistical Natural Language Processing
,
1992,
COLING.
[3]
J. Hammersley,et al.
Monte Carlo Methods
,
1965
.
[4]
George R. Doddington,et al.
The ATIS Spoken Language Systems Pilot Corpus
,
1990,
HLT.
[5]
Yves Schabes,et al.
Stochastic Lexicalized Tree-adjoining Grammars
,
1992,
COLING.
[6]
Rens Bod,et al.
A Computational Model of Language Performance: Data Oriented Parsing
,
1992,
COLING.
[7]
Frederick Jelinek,et al.
Basic Methods of Probabilistic Context Free Grammars
,
1992
.