Mobile advertisement click fraud detection method, system, terminal and medium

Abstract

The invention provides a mobile advertisement click fraud detection method and system, which converts click events for advertisements in an app into a data dependency graph, and then into a feature vector, reconstructs it through a variational autoencoder, and uses the reconstruction method. The error determines whether the click event is a click fraud event, and locates the code position where the click fraud occurs, so that the relevant personnel can investigate and collect evidence on the app click fraud. At the same time, a click fraud detection system and a terminal for performing the above-mentioned static detection-based mobile advertisement click fraud detection method are provided. At the same time, a corresponding terminal and medium are provided. The present invention utilizes static detection technology and one class classification algorithm, does not require manual intervention, and does not require a large number of labeled click fraud samples in the early stage, can realize more efficient, fast and flexible click fraud detection, and meet the needs of related industries.

Description

technical field [0001] The present invention relates to the technical field of static detection, and in particular, to a method, system, terminal and medium for detecting click fraud of mobile advertisements based on static detection. Background technique [0002] For decades, with the unprecedented popularity of smartphones, pay-per-click (PPC) mechanisms have been widely deployed to incentivize app developers to embed advertisers' ads in their apps, where advertisers are viewed by users based on embedded ads. The developer is paid for the number of clicks. However, PPC mechanisms also face the growing threat of click fraud. By employing click fraud tactics, some developers use code to generate "fake" ad click events that do not come from real users in order to generate additional revenue from ad networks. For example, attackers can embed malicious code on fraudulent apps or third-party SDKs that are exploited by other unsuspecting app developers and can automatically tri...

AI Technical Summary

Problems solved by technology

[0007] 1. No appropriate features are defined at the decompiled code level, so the difference between fake click events and normal click events cannot be reflected

[0008] 2. These features cannot be extracted automatically

[0009] 3. It is impossible to accurately distinguish between fake click events and normal click events under the premise of limited understanding of malicious samples

[0010] At present, there is no description or report of the similar technology of the present invention, and no similar data at home and abroad have been collected yet.

Images