USING MULTIPLE DATA SOURCE TO ENHANCE PART-OF-SPEECH TAGGING FOR CHINESE PATENT