A Distributed Approach for Spam Collection and Detection Using Honeypots

Junk mail or spam is one of the serious problems of users in their daily electronic communications. Low working efficiency, disk space occupation, and phishing attacks are justsome spam consequences. With the advent of spam, the researchers have attemted at introduce methods to deal with thisphenomenon. Text-based Spam filtering can be considered as a text categorization problem. The most majority works in thisarea are dedicated to machine learning-based techniques. Despite the excellent results, machine learning algorithms are time consuming and highly dependent on the training data. In thispaper a method has been proposed for distributed spam detection base on honeypot traps and header features. Thisapproach obtains fresh training data through hybrid honeypot which deceives spammers in fidelity manner without any footprint. The accuracy of spam detection is high while using only header features along distributed classifier based on Naive Bayes algorithm. The results show increasing the number of classifier lead to improve accuracy measures like Recall and Precision.

کلیدواژه ها:

Spam ، Honeypot ، machine learning ، Naïve Bayes ، Honeyd