Study on Spam SMS Detection Based on TF-IDF and Random Forest

With the proliferation of mobile internet, SMS, as a fundamental communication modality, remains widely adopted. However, the rampant dissemination of spam SMS severely undermines user experience and information security. Traditional rule-based keyword filtering approaches exhibit limited flexibility, struggling to adapt to the evolving patterns of spam SMS. This study proposes a robust spam SMS detection method integrating TF-IDF and random forest algorithms. Specifically, TF-IDF is employed to extract textual features from SMS content, converting unstructured text into numerical feature vectors that quantify the semantic importance of terms. These vectors are then input into a random forest classifier to construct a discriminative model capable of distinguishing spam from legitimate messages. Experimental validation on a publicly available Kaggle dataset demonstrates the method's superior performance, achieving an accuracy of 97.73 %. The results highlight that the proposed framework offers an efficient and reliable solution for automated spam SMS recognition, effectively addressing the limitations of conventional rule-based methods and providing a scalable approach for real-world deployment.