Co_NBM: A Semi-Supervised Categorization Algorithm Based TEF_WA Technique

We propose a semi-supervised categorization algorithm Co_NBM incorporating co-training and TEF_WA technique. General co-training algorithm relies on the assumption that the features set can be split into two compatible and independent views. However, the assumption is usually violated to some degree in practice and sometimes the natural feature split does not exist. TEF_WA technique utilizes term evaluation functions to reduce dimensionality and adjust terms weight. Now, it is used to construct multiple views. Our experimental results show that utilizing unlabeled data Co_NBM can significantly decrease classification error, especially when labeled training data are sparse.