Enhancing Fake Profile Detection Through Supervised and Hybrid Machine Learning - A Comparative Analysis

Aug 3, 2026

Social networks generate large volumes of user-generated content, which also enables fraudulent actors to create fake profiles that spread misinformation, conduct deception, and threaten genuine users’ privacy. This study presents a comparative approach to fake profile detection using machine learning, combining unsupervised clustering (K-means and K-medoids) with supervised classifiers. Models such as KNN, SVM, Bernoulli Naïve Bayes, logistic regression, and linear SVC are trained on analyzed behaviors, engagement metrics, and content features to capture patterns distinguishing real from deceptive accounts.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages12

Reading Time30 min

AccessFree to download

Introduction

Problem of fake profiles in social networks

Motivation for supervised–unsupervised hybrid modeling

Studies and Method

Related studies

Detection methods and feature distinctions

Enhancing fake profile detection through supervised and hybrid machine learning: a comparative analysis Ismail Bensassi, Oussama Ndama, Mohamed Kouissi, El Mokhtar En-Naimi Faculty of Science and Technology of Tangier, Abdelmalek Essaadi University of Tetuan, Tangier, Morocco | Article history:
Received Jun 28, 2024 Revised Aug 28, 2024 Accepted Sep 2, 2024 | | --- | | Keywords:
Bernoulli Naïve Bayes Fake profiles detection K-medoids
Linear SVC and K-means Logistic regression Supervised and unsupervised machine learning algorithms as KNN, SVM
User behavior | Corresponding Author: In modern times, social networks have become ubiquitous platforms facilitating widespread information dissemination, resulting in significant daily data generation. This increase in data production encompasses a wide range of user-generated content, which in turn promotes the proliferation of fraudulent users creating fake profiles and engaging in deceptive activities. This article aims to address this challenge by employing machine learning algorithms to accurately identify fake profiles. The research involves a thorough analysis of various user behaviors, engagement metrics, and content attributes within social platforms. The primary goal is to develop robust models capable of effectively detecting deceptive profiles by meticulously examining user activities and content characteristics. The study explores the application of robust methodologies such as K-means and K-medoids clustering, alongside supervised machine learning classifiers including K-nearest neighbors (KNN), support vector machine (SVM), Bernoulli Naïve Bayes (NB), logistic regression, and linear support vector classification (SVC), specifically tailored for the detection of fake profiles. This is an open access article under the CC BY-SA license. Ismail Bensassi Faculty of Science and Technology of Tangier, Abdelmalek Essaadi University of Tetuan B.P 416, postal code: 90000, Tangier, Morocco [Email: bensassi.ismail@gmail.com](Email: bensassi.ismail@gmail.com) Article Info ABSTRACT 1. INTRODUCTION The rapid expansion of social media platforms in recent years has profoundly impacted various societal realms, particularly in marketing and celebrity self-promotion, which have significantly broadened their follower and fan bases. While these platforms offer extensive benefits such as widespread information dissemination, they also introduce challenges related to privacy, misinformation, and online security. A critical issue among these challenges is the widespread presence of fake profiles, which pose risks by disseminating misinformation, engaging in fraudulent activities, and compromising the privacy of genuine users. Addressing the complex challenges posed by fake profiles requires a comprehensive approach integrating technological solutions aimed at detecting, preventing, and mitigating the negative impacts associated with these deceptive entities. This study focuses on harnessing machine learning algorithms to accurately detect fake profiles on social networks. By meticulously analyzing user behaviors , engagement metrics, and content characteristics within these platforms, the goal is to develop robust models capable of effectively identifying deceptive profiles. Existing studies are carried out with supervised algorithms which are generally more accurate for specific tasks with labeled data, while unsupervised algorithms are better for discovering hidden structures in unlabeled data. In our approach, we have opted for a mixture of both types of algorithm, which allows us to take advantage of the strengths of each method, using unsupervised algorithms to explore and prepare the data before applying supervised methods for accurate predictions. This combination is often used to optimize performance in situations where labeled data is scarce or expensive to obtain, and also to get the best results in terms of time and accuracy in detecting false pro

FAQs

What problem does the study address?

The study addresses the detection of fake profiles on social networks, which enable misinformation, fraudulent activities, and privacy compromise for legitimate users.

Which machine learning methods are used in the proposed approach?

It combines clustering methods (K-means and K-medoids) with supervised classifiers including KNN, SVM, Bernoulli Naïve Bayes, logistic regression, and linear SVC.

What types of signals/features are analyzed to distinguish fake from genuine profiles?

The study analyzes user behaviors, engagement metrics, and content characteristics, focusing on features associated with user activity and profile statistics such as follower/friend/status-related counts.

Aug 3, 2026

Upload by Clementine