SMS Spam Detection Using Machine Learning Approach

Abstract Currently, as the popularity of mobile phones has increased, Short Message Service (SMS) has grown tremendously. The minimal cost of messaging services has increased spam or unsolicited messages sent to mobile phones. There are differences between spam filtering for text messages and emails. Emails have a set of big datasets, while the actual databases for SMS spam are very limited. Because of the small size of text messages, the features used for classification are smaller than the equivalent number in emails. Text messages consist of abbreviations and have less formal language than that of emails. Short Message Services (SMS) spam has become a pressing issue in mobile communication, disrupting user experiences and posing privacy threats. This study develops a useful system for identifying spam messages in SMS communications. It presents a machine learning-based framework for detecting SMS spam, utilizing a Multi Layer classifier. This is aimed at tackling the problem of spam messages in SMS communications through the development of a robust and efficient spam detection system. This entails a data preprocessing procedure to prepare the raw SMS dataset. The TF-IDF technique was used to handle feature extraction to represent the text data numerically. This enables the model to capture relevant characteristics distinguishing spam from non-spam messages. The model was trained using the preprocessed data and evaluated through cross-validation. The results highlight the scalability and reliability of this approach, providing a practical solution for enhancing SMS spam detection systems and improving user security in mobile communication, employing the multi-layer classifier for an effective spam detection system, ensuring the models' optimal performance while preventing overfitting to deliver a comprehensive solution to the persistent issue of SMS spam.