A Meta Classifier Model for SMS Spam Detection using MultinomialNB - LinearSVC Algorithms

There has been a rise in the use of mobile devices during the past few decades, networks and short message services (SMS) has emerged as a form of communication. SMS spam is a problem for SMS users as well. SMS spam, also referred to as bulk texts, is any irrelevant communications sent via mobile networks [2]. Several factors contribute to the abundance of spam communications, The fact that so many individuals use mobile phones increases the probability that they will be the object to send bulk messages which is known as spam [1]. Furthermore, sending spam is inexpensive, which may be great news to the attackers. Particularly, spam detection is a highly advanced research area with a variety of established algorithms [15]. In order to correctly identify spam data or communications, this approach looks into a Multinomial Naive Bayes-Linear SVC methodology [4]. Pre-processing is applied to the input dataset to get rid of characters and content that aren’t appropriate or relevant [5]. To train the model a Multinomial Naive Bayes-Linear SVC methodology is used for the prediction of spam messages [10]. The final test outcomes have demonstrated that Multinomial Naive Bayes-Linear SVC model works better than earlier models in spam detection such as LSTM, SVM and naive bayes, with an accuracy of 9S.3S%.