Analysis of electronic mail, comparison of methods fuzzy logic and Naive Bayesian

One of the main facilities that internet technologies provide is accepted as communication via e-mail. But within the rise in users, increase in the number of e-mails, new methods of earnings in e-mails and electronic advertisements, a new definition ?SPAM? came into the picture. The irresistible rise in SPAM e-mails started disturbing the users in a serious manner. Some of the main troubles can be listed as waste of time for users, disturbing content, letting fakery via internet and for causing capacity problems for service providers. Common researches show that SPAM e-mails are definitely unwanted by the users.Therefore, seperating the SPAM e-mails from the other ones became an absolute necessity in the internet society. Because the number of the SPAM e-mails are considerably high, some artificial intelligence methods are tested before. There is a need to keep on analyzing these methods since the performance, capabilities and resolution of the methods are exteremely important.In this thesis, a research of seperating SPAM e-mails from the standart e-mails is completed. The two methods for separating e-mails form SPAM e-mails, Naive Bayesian and Fuzzy Logic are investigated. First, a virtual e-mail server provider is generated by a simulator application and sending e-mails by the virtual users are carried out. After that, the analysis are made among a chosen user. The e-mails that are recieved by that user are used to generate a sample data classification chart. By using this chart, a SPAM probability table is derived. The important characteristic of this method is that the SPAM probabilities are derived among the e-mails that are recieved by the user before. Two different SPAM states are reviewed fort he two different methods. By evaluating the datas obtained by the application results, Naive Bayesian method can be seen as more efficient in seperating the SPAM e-mails.