SUBJECT: &NAME results Hi &NAME , Here are the baseline results for my classifier on the lingspam corpus : precision for GENUINE : &NUM recall for GENUINE : &NUM precision for &NAME : &NUM recall for &NAME : &NUM F1 for GENUINE : &NUM F1 for &NAME : &NUM There seems to be some confusion in the literature about the exact definitions of precision and recall for the 2-class classification problem . I define recall as &NUM ( &NAME + FN ) and precision as &NUM ( &NAME + FP ) where &NAME true positive &NAME false positive &NAME false negative i.e. recall is measuring the number of correctly classified units for class &CHAR against the number missed , whereas precision is measuring the number of correctly classified units for class &CHAR against the total number of class &CHAR classifications made . In some of the literature these measurements are the other way around ! This gives a precision / recall / F1 for each class . &NAME