Spam filtering using data mining methods

The importance of e-mail communication in our lives has continually increased since the commerce is developed over internet channels, and there is fast and economic communication. Very low operation cost provides transferring a large number of data within a few seconds over long distances.Sending a large number of copies of the same message stringently to the people who are not willing to receive over the internet is called spam.UCE (Unsolicited Commercial e-mail) and UBE (Unsolicited Bulk e-mail) which are kinds of spam messages sent via e-mail, as it can be inferred from the names, are introductory e-mails which is actually undesirable.There is not an available unique technique or an available solution combined by the techniques in which the problem of undesirable e-mail is solved. There have been lots of data mining approaches aimed at determining unsolicited e-mails.Data mining is the process of finding the interesting patterns which are obviously not part of the data. In spam filtering, there are two kinds of approaches. One is filtering by constructing the rules by knowledge engineering. Second is classification within datasets prearranged via the techniques known as data mining separated from machine learning by applying machine learning techniques over very large datasets.Within the scope of this thesis, spam filtering has been implemented by applying data mining techniques over attribute space model formed on the basis of e-mail datasets.