Spam Detection System Based on Hybrid Scoring
Stefka Popova, Hristo Nenov, Donika Stoyanova · 2024
Spam continues to be a serious problem for the vastly growing online communications. The is the reason for ongoing research and development of various solutions and systems for spam prevention and detection. This paper presents a spam detection system based on hybrid scoring which uses machine learning algorithms to analyze given email and classified as either spam or ham. Proposed system uses TF-IDF in the process of feature selection and works for emails written both in English and Bulgarian. Algorithms as Naïve Bayes, Support Vector Machines, Logistic Regression, Decision Tree, and Random Forest are implemented and tuned, and numerous experiments are conducted. Support Vector Machines and Random Forest models achieve best results in terms of classification accuracy, reducing classification error to less than 2%.