4th International Conference on Spoken Language Processing

Philadelphia, PA, USA
October 3-6, 1996

Estimating Markov Model Structures

Thorsten Brants

Universität des Saarlandes, Computational Linguistics, Saarbrücken, Germany

We investigate the derivation of Markov model structures from text corpora. The structure of a Markov model is its number of states plus the set of outputs and transitions with non-zero probability. The domain of the investigated models is part-of-speech tagging. Our investigations concern two methods to derive Markov models and their structures. Both are able to form categories and allow words to belong to more than one of them. The first method is model merging, which starts with a large and corpus-specific model and successively merges states to generate smaller and more general models. The second method is model splitting, which is the inverse procedure and starts with a small and general model. States are successively split to generate larger and more specific models. In an experiment, we show that the combination of these techniques yields tagging accuracies that are at least equivalent to those of standard approaches.

Full Paper

Bibliographic reference.  Brants, Thorsten (1996): "Estimating Markov model structures", In ICSLP-1996, 893-896.