<![CDATA[Phishing email attacks are a serious threat in the digital ecosystem because they can trick users into leaking sensitive information or accessing malicious links. This study aims to develop a phishing email classification model based on the Support Vector Machine (SVM) algorithm combined with Natural Language Processing (NLP) techniques to improve detection accuracy. The process begins with the tokenization, text cleansing, and feature extraction stages using the TF-IDF approach, which is further used as input into the classification model. Various SVM kernels, including linear, radial basis function (RBF), and polynomial, are tested through the grid search method with parameter tuning such as C, gamma, and degree. The results showed that SVMs with polynomial kernels produced the highest accuracy of 97.85%, surpassing other algorithms such as Naïve Bayes, Random Forest, and Logistic Regression. These findings indicate that the integration of NLP and SVM with proper parameter tuning provides an effective solution in mitigating phishing email attacks. This model can be the foundation for the development of a more adaptive and efficient cybersecurity system.]]>
