engleski

Hidden Markov Model (HMM) for tagging of Croatian language texts

Encouraged by the recent results in creating automatic taggers for seven different European languages, an attempt has been made to create one for the Croatian language as well. In the paper, the following aspect will be presented: the results achieved in the framework of the ESPRIT-291/860 project "Linguistic Analysis of the European Languages", past development in this area for Croatian language, characteristics of Croatian language, description of the corpora available at the moment, short introduction of the HMM as the mathematical background for the future research and the results.

POS; Part-of-Speech; stochastic automatic triggers; HMM

nije evidentirano