Spam Classification Using MOEA/D


In mathematics, it’s very easy to find the maximum point or minimum point of a function or a set of functions, but it’s difficult to find a set of function simultaneously in the real world due to the different kinds of mathematical relationships between objective functions. So the multi objective optimization algorithm has the ability to deal with a many objectives instead of one objective, because of the difficulties in the classical methods of multi objectives optimization, the evolutionary algorithm (EA) is effective to eliminate these difficulties, in order to apply the evolutionary algorithms to improve the multi-objective optimization algorithm, the multi - objective evolutionary algorithm based on decomposition is one of the algorithms that solve multi objective optimization problems. This paper aims to enhance the e-mail spam filtering by using multi - objective evolutionary algorithm for classifying the e-mail messages to spam or non-spam in high accuracy. The first step in the proposal is applying normalization. The second step is applying feature selection which is implemented to choose the best features. Finally, implement multi - objective evolutionary algorithm based on decomposition. The evaluation of the performance of model by using testing databases from the spam database. The model depended accuracy as a criterion to evaluate model performance. The experimental results showed that the proposed system provides good accuracy in the experiment 1 (91%), very good accuracy in the experiment 2 (92%) and excellent accuracy in the experience 3 (98%).