Abstract-This approach is highly effective in creating predictions, especially when compared to certain traditional forms of supervised learning. When using ASSD, identifying individuals within a group becomes simpler and requires less effort. To update the spammer detection model without requiring retraining of the model's users, incremental learning is used as a technique because social scammers often modify their behavior to deceive the spammer detection model. When benchmarked against other controlled and semi-supervised machine learning algorithms, the Social Honeypot Dataset is used to compare ASSD's performance. The study's findings suggest that the proposed model outperforms baseline approaches in terms of memory capacity and accuracy. Additionally, ASSD maintains its high accuracy in identifying spammers by continuously updating its model with newly collected data from social media. 

| DOI: 10.17148/IARJSET.2023.10834