Introduction of rules into a stochastic approach for language modelling
Automatic morpho-syntactic tagging is an area where statistical approaches have been more successful than rule-based methods. However, the currently available systems based on statistical algorithms appear to be unable to handle span dependencies and to model infrequent structures. On the other hand, there exist strong arguments that the weakness of statistical techniques may be compensated by rule based methods. The main purpose of the present paper is to take advantage of the features that are complementary in the two approaches, and to propose a hybrid model for the design of an improved tagger, called ECSta. The proposed hybrid strategy for jointly using a set of rules and a stochastic model has several positive characteristics: (a) The application of the rules during the probabilistic process inhibits the error propagation. Such an improvement could not be obtained by a post-processing analysis. (b) The ECSta tagger combines, in a consistent manner, a stack-decoding algorithm with the Viterbi classical one. (c) The main interest pro-vided by the proposed approach consists in introducing under specified structures and log span dependencies in a stochastic model. For the entire collection see [Zbl 0927.00044].
This page was built for publication: Introduction of rules into a stochastic approach for language modelling
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4942756)
