An algorithm for text categorization
Anestis Gkanogiannis, Theodore Kalamboukis · 2008
A novel and efficient learning algorithm is proposed for the binary linear classification problem. The algorithm is trained using the Rocchio's relevance feedback technique and builds a classifier by the intermediate hyperplane of two common tangent hyperplanes for the given category and its complement. Experimental results presented are very encouraging and justify the need for further research.