SomaliPhishingDataset v1.0

SomaliPhishingDataset is a Somali-language cybersecurity dataset for binary classification of digital messages as phishing (1) or ham/benign (0). The public release contains 8,684 messages collected from Whatsapp, Telegram, SMS, and Facebook/Tiktok. It includes binary labels and a rule-based annotation scheme (Rule1–Rule5) capturing common phishing tactics. All public text is PII-masked using tokens such as [URL], [PHONE], [EMAIL], and [ID]. This dataset supports reproducible benchmarking and phishing detection research for a low-resource language setting.