Towards a Machine Learning Approach for Detecting Click Fraud in Mobile Advertizing

Abstract

In recent years, mobile advertising has gained popularity as a mean for publishers to monetize their free applications. One of the main concerns in the in-app advertising industry is the popular attack known as “click fraud”, which is the act of clicking on an ad, not because of interest in this ad, but rather as a way to generate illegal revenues for the application publisher. Many studies evaluated click fraud attacks in the literature, and some proposed solutions to detect it. In this paper, we propose a click fraud detection model, hereafter CFC, to classify fraudulent clicks by adopting some features and then testing using KNN, ANN and SVM. In fact, based on our experimental results, the different featured classifiers reached an accuracy higher than 93%.