Linguistic Evaluation in the Classification in Portuguese Texts

This paper evaluates the performance of support vector machines, Naive Bayes, and neural networks as classifiers for the categorization of Portuguese texts. We present several experiments with two different corpora with different feature selection strategies. We consider the use of linguistic information in the definition of grammatical groups. A comparison of classifiers is presented and the error margins show excellent results when using a specific feature selection in association with the right classifier.