The focus of this dissertation is the development and validation of a novel method for supervised text classification to be used effectively when small sized training sets are available. The proposed approach, which relies ...