Capturing observations in a nonstationary hidden Markov model

This paper is concerned with the problem of morphological ambiguities using a Markov process. The problem here is to estimate interferent solutions that might be derived from a morphological analysis. We start by using a Markov chain with one long sequence of transitions. In this model the states are the morphological features and a sequence correponds to a transition from one feature to another. After having observed an inadequacy of this model, one will explore a nonstationary hidden Markov process. Among the main advantages of this latter model we have the possibility to assign a type to a text, given some training samples. Therefore, a recognition of “style” or a creation of a new one might be developped.