SUBJECT: Re : &NAME tuple representation of Reuters-21578 Hi , I my knowledge , following paper use itemset for text categorization . [ &NUM ] &NAME &NAME , &NAME &NAME , &NAME &NAME , &NAME &NAME , &NAME association-based text classification , in Proc . 9th &NAME &NAME &NAME . Information and Knowledge Management ( CIKM' &NUM ) , pages &NUM , &NAME , &NAME , &NUM . [ &NUM ] &NAME &NAME , &NAME &NAME &NAME , &NAME &NAME &NAME by &NAME Association , in Proceedings of the &NAME International Conference on Data &NAME ( ICDM' &NUM ) , &NUM . In a study of mine , we have observed an improved performance of &NAME classifier , and c4.5 classifier while taken itemset as features for text classifier . &NAME &NAME <QUOTE> Dear &NAME , we are planning to do research on the comparison of decision rule classifiers on various data sets . This should also include text categorization tasks . Does anyone of you provide a feature tuple reresentation of the Reuters-21578 &NAME training and testing sets , which could be used for such experiments ? Having such an input available would be a great support for us ! Best regards , &NAME &NAME &NAME &NAME &NAME University of &NAME , Center for &NAME &NAME &NAME &NUM , Room &NUM , D-28359 &NAME &NAME &NAME &NUM &NUM , D-28334 &NAME , &NAME &NAME : &NUM &NAME : &NUM &NAME : &EMAIL www : &WEBSITE Currently at : &NAME - &NAME &NAME and Inference Laboratory &NAME &NAME University , &NAME &NAME &NUM University Drive , &NAME , VA &NUM , &NAME &NAME : &NUM <END_QUOTE>