SpamDam: Enabling Privacy-Preserving Discovery and Analysis of SMS Spam
Year: 2026
Type: conference-paper
Abstract: In this study, we introduce SpamDam, a machine learning toolchain designed to overcome key challenges in understanding and mitigating SMS spam, especially the unavailability of up-to-date and global-wide SMS spam datasets, and the increasing privacy concerns that can be incurred in both collecting SMS messages and training SMS spam detection models. SpamDam comprises two modules: A radar that discovers SMS spam messages as reported by victims on online social networks; An inspector that statistically analyzes and gains insights from SMS spam messages. Leveraging SpamDam, we have discovered from Twitter and Weibo, over 76K distinct SMS spam messages that were reported by victims between 2018 and 2023, giving the largest SMS spam dataset.
Author: Yekai Li
Institution: University of Science and Technology of China
Language: English
Topic: Spam and Phishing Detection
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7225167659