A fuzzy expert system approach for spam detection

This paper attempts at providing a fuzzy expert system to detect spams. By definition, any unsolicited e-mail to user is called spam. In this respect, there is no definite criterion to distinguish legitimate e-mails from spam, hence a fuzzy expert system is proposed to address above problem. Our proposed approach is to use various e-mail features to drive a fuzzy model, which is then an expert system and defuzzification is used to bring the results. The criteria Recall and Precision were applied to measure the system and the best values obtained were 97.4% and 99.3% respectively. The accuracy here can compete with other methods.