The massive use of e-mail makes text categorization once more an
important issue as everyone would like e-mail accounts that could
prevent undesired mail successfully. Proposed method is intended
to classify incoming e-mails either as unsolicited e-mails (spam)
or good e-mails (ham). This prediction is based on a machine
learning engine; the keywords for each one of the two categories
are retained in a training stage and then used to predict the
category to which new e-mails should belong, in a test stage.