HYBRID WATER CYCLE OPTIMIZATION ALGORITHM WITH SIMULATED ANNEALING FOR SPAM EMAIL DETECTION

Abstract

Spam is referred to unsolicited commercial e-mail from someone trying to give some information that the receiver did not expected. This kind of email usually defined as junk and unwanted. As a filtering step, the spam email filters are implemented in conjunction to reduce this type of e-mails. Unfortunately, new spam email attributes have caused the spam email filter characteristic insufficient and inefficient to handle the large amount of email. This problem is due to the large number of features that the spam classifier needs to evaluate. By the help of feature selection method, the number of features can be reduced. However, the optimal number of features remains a problem and requires further investigation. In this thesis, a new hybrid method has been introduced to make the spam email feature selection more accurate by using the metaheuristic feature selection optimization approach. The proposed method is based on the hybridization of Water Cycle Algorithm with the Simulated Annealing to optimize the results. This study used a methodology that included groundwork, induction, improvement, assessment, and comparison quality. For the training and validation datasets, cross-validation was performed, and seven datasets were used to evaluate the suggested spam classification