Among the anti-spam techniques exist in the literature include those based on machine learning and those not based on machine learning (c.f. Figure 1):
SPAM Filtering approaches
There are two types of approaches for the detection and the filtering of SPAM (see Figure 1): the approach that is not based on Machine Learning and the Machine Learning-Based Approach.
This approach consists of using one of the techniques or strategies of detection and filtering of SPAM as analysis of contents, black-lists, white-lists and authentication of a mailbox or a heuristic and meta-heuristics.
Content analysis is an application implemented on the mail server as a complement to the user's mail application. Its role is to give a probability to an email to be a SPAM (or HAM) according to its contents.
The block list contains viral email’s addresses so that we should block any email that is sent from one of these addresses. It is filled and modified by the end who are not enough competent in computing will not be able to use this technique effectively and appropriately.
The principles of the legal computing to verify the signature of the email's author (authenticate the physical or real sender) and to trace an email to detect proxies and diversions of routing which the spammers use to bypass the other techniques of detection and filtering of SPAM.
Many of meta-heuristics have been proposed. Among these remarkable works, we can quote the Modelling of immune systems for the detection and the filtering of SPAM, meta-heuristic based on Artificial neural network with three layers, social bees for the filtering of the SPAM.