Year: 2025
Type: article
Abstract: The growing reliance on social networks like Facebook, Instagram, Twitter, and LinkedIn has made them central to modern communication and business, but it has also introduced threats from fake profiles. These accounts are created to mislead users, spread spam, conduct scams, or manipulate public opinion, undermining trust, privacy, and safety. Traditional manual or rule-based detection methods are ineffective due to the sheer volume of accounts and the adaptability of attackers. This research presents a machine learning framework for detecting fake profiles across multiple platforms. Logistic Regression and Random Forest algorithms are trained on datasets containing both genuine and fake accounts. Classification uses features such as followers-to-following ratio, bio completeness, profile picture presence, posting frequency, and account age. Results show that Random Forest consistently outperforms Logistic Regression, achieving accuracy above 90% on all platforms. The study demonstrates the effectiveness of machine learning in improving the safety and trustworthiness of social networks and provides a foundation for future extensions using more advanced detection methods.
Author: Soumya Munji
Language: English
Topic: Spam and Phishing Detection
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4415030137