Abstract: The misjudgement of spam has always been the difficulty in the anti-spam area. Experiments based on Tsinghua university E-mail systems and anti-spam gateway system(Sep,2011–Oct,2012) were deployed and made, analyzing the history of recalling spam behavior of mail users, and using spam text similarity calculation and intelligent key parameters prediction method to predict the user's interest in the current pending spam, which can help users automatically recall their potential interested spams which were misjudged, based on the users' subjective choices and experience, solving the problem that cannot be eliminated by traditional filtering methods.