Propose hybrid ACO and NB to enhance spam filtering system

Abstract EN

Unwanted e-mails became one of the most risk experienced by e-mail users, which may be either harmless or e-mails that represent a threat to the internet.

Filtering systems are used to filter e-mail messages from spam.

This paper introduces a proposed hybrid system to filter the spam; the proposal hybrid Ant Colony System (ACS) and Naive Bayesian (NB) classifier.

Where, ACS will depend on the Information Gain (IG) as a heuristic measure to guide the ants search to select the optimal worst features then omitting these features.

The remind features will be the subset which is used to train and test NB classifier to classify whether the mail message spam or not.

The proposal is experimented on spambase dataset, and the results show that; the accuracy, precision and recall with NB which use a subset of features extracted by proposing IG-based ACS is higher than the traditional NB with all set of features.