Improved Spam Detection Using DBSCAN and Advanced Digest Algorithm

Abstract

Mail is one of the most popular and frequently used ways of communication due to its worldwide accessibility, relatively fast message transfer, and low sending cost. Nowadays, detecting and filtering are still the most feasible ways of fighting spam mails. There are many reasonably successful spam mail filters in operation. The identification of spam plays an important role in current anti-spam mechanism. For improving the accuracy of spam detection, we present an improved Filtering technique which is based on the Improved Digest algorithm and DBSCAN clustering algorithm. Using this technique, mails are represented using improved digest algorithm and then clustered using DBSCAN clustering algorithm. All similar mails which always categorized as spam are identified and clustered together where good mails that don’t look similar like other mails are not clustered. This method greatly improves the filtering accuracy against latest proposed algorithms by 30 % and improves the resistance of spam detection against increased obfuscation effort by spammers, while keeping miss detection of good mails at a similar level of older filtering methods.