Cyber-Harassment Among Sri Lankan Female Undergraduates: Prevalence, Patterns, and AI Detection

Abstract

Cyber harassment has emerged as a serious concern for university students, especially female undergraduates, who are increasingly exposed to gender based abuse in online spaces. Despite this growing issue, there is limited empirical evidence in Sri Lanka on how female students experience cyber harassment, how they respond to it, and how such data can be used to inform future detection systems. This study aimed to examine the prevalence, patterns, and perceived effectiveness of responses to cyber-harassment among Sri Lankan female undergraduates and generate an empirical foundation for future AI-based detection. A sequential exploratory mixed-methods design was employed. Qualitative data were collected through direct field visits and in person interviews with ten female undergraduates to explore their lived experiences, emotional impact, and coping responses. The insights gained were then supported through a structured Google Forms survey administered to 150 female undergraduates. The survey examined exposure to cyber harassment, the platforms involved, the identity of perpetrators, awareness of safety features, and trust in institutional support. The findings revealed that 73% of respondents had experienced cyber-harassment at least once, with WhatsApp being the most frequently reported platform, followed by Facebook, Instagram, and TikTok. Most incidents involved strangers or unidentified accounts, indicating the role of anonymity in online abuse. Although most participants were aware of blocking and reporting tools, very few found them effective, and trust in institutional support remained low. The interview data reinforced these results by highlighting emotional distress, reduced confidence, and reluctance to remain active online. The study concludes that cyber-harassment is a persistent and significant issue for Sri Lankan female undergraduates, and its findings provide a useful basis for developing an AI-assisted detection system to support early identification and intervention of such incidents.