Bilingual Spam SMS detection using Machine Learning
Babe Sultana, Zakia Afrin, Farhana Ryhan Kabir, Dewan Md. Farid · 2023
Every day, Bangladeshis receive several SMS spam messages on their phones in both Bangla and English text. Spam is defined as unsolicited bulk communications in a variety of formats, such as unsolicited advertisements, credit offers, and fake lottery prize messages. The daily volume of (SMS) traffic is consistently increasing. Consequently, instances of mobile attacks, such as spammers bombarding the service with unsolicited messages sent to groups of recipients, are also rising significantly. Even with the filtering systems in place, mobile spam is a developing problem as the number of spam messages increases daily. Due to the complexity of the messages generated by spammers, spam classification has become increasingly challenging. Previously, Bangla and English SMS spam detection was done separately, but Bangladeshi people must detect Bangla and English spam SMS at the same time. In this research, we developed a bilingual dataset by combining our own Bangla dataset with an online-accessible English dataset. We then used supervised machine learning algorithms for detecting Spam messages (SMS). Based on findings from experiments, every algorithm provides greater accuracy and among them SVM performs better with 97.89% accuracy.