Research And Implementation Of A Content-Based Cross-Platform Detection Method For Advertising Fraud

In recent years,the rapid development of the Internet has led to the advent of the digital age.Due to the unique advantages of the Internet,such as digitalization,wide dissemination,and multiple formats,many advertisers have turned their advertisements from traditional forms to the Internet.However,in order to increase the click-through rate of advertisements for greater benefits,some advertisers and developers use various methods to induce or even deceive users to click on advertisements,which has caused a large number of security problems in ad fraud.Compared with other types of advertising fraud,content-based advertising fraud(Visual Fraud)has a special form and is common on multiple network platforms.Therefore,this article focuses on content-based advertising fraud on the Web and mobile terminals.The design of advertising content deceives users,causing users to misunderstand the advertisements and causing users to misuse and click.At present,the research on advertising fraud mainly focuses on click fraud on the web side,that is,increasing the click-through rate of advertisements through non-manual false clicks;however,the mobile side mainly focuses on the irregular behavior of mobile advertising(for example,frequent pop-ups)and its static properties(for example,large area advertising,large number of advertising).However,content-based advertising fraud(Visual Fraud)has been widely spread in network platforms,but the existing research only focuses on the text of the advertisement and the content of the advertisement,but not the other content of the advertisement(for example,images).Based on the existing research on content-based advertising fraud(Visual Fraud),this paper designs and implements a set of effective content-based advertising fraud cross-platform detection methods.The main results of this paper are as follows:1)Implementation of mobile and Web Advertisement extraction;2)By investigating the national advertising law and the advertising specifications issued by various e-commerce platforms,as well as manually identifying and evaluating existing advertising resources,summarize several common types of content-based advertising fraud(Visual Fraud)to establish a classification method,for each type of image feature,use deep learning and other technologies to develop specific detection method;3)For extended analysis,this article analyzes 100,000 advertising examples crawled from Internet advertising collection platforms and performs ad fraud detection,and then a simple analysis of their advertisers and advertising platforms,while horizontally comparing the occurrence of advertising fraud on mobile and web,and draw some preliminary conclusions.