T-HMM: A novel biomedical text classifier based on hidden markov models

Abstract

In this paper, we propose an original model for the classification of biomedical texts stored in large document corpora. The model classifies scientific documents according to their content using information retrieval techniques and Hidden Markov Models. To demonstrate the efficiency of the model, we present a set of experiments which have been performed on OHSUMED biomedical corpus, a subset of the MEDLINE database, and the Allele and GO TREC corpora. Our classifier is also compared with Naive Bayes, k-NN and SVM techniques. Experiments illustrate the effectiveness of the proposed approach. Results show that the model is comparable to the SVM technique in the classification of biomedical texts.