Application Research Of Network Advertising Fraud Detection System Based On Xgboost

With the rapid development of information technology,the scale of the online advertising industry continues to expand,and the problem of online advertising fraud has gradually expanded and exposed with the development of the industry.Online advertising fraud refers to the use of machines or artificially imitating the behavior of normal network users to access online advertisements for the purpose of obtaining commercial profits.Online advertising fraud not only caused direct economic losses for advertisers,but also caused a huge credibility crisis for the online advertising industry.Solving the problem of online advertising fraud is a top priority for the development of the online advertising industry.This paper combing the theories and methods accumulated in the research of preventing,detecting and restraining the Internet advertisement fraud research.Based on the research methods and conclusions of predecessors'online advertisement click fraud detection under CPC settlement mode,Combined with the characteristics of QD company's advertising platform business model and CPA settlement method,this paper proposes an advertising effectiveness fraud detection method for QD company's online advertising platform business.The network advertisement fraud detection system based on XGBoost algorithm is realized by combining the advertising effect data with the user behavior data for data collection and feature extraction.Through the test and effect evaluation of QD company's actual online business data,the feasibility and effectiveness of XGBoost algorithm combined with CPA settlement method feature data extraction in network advertisement fraud detection is verified.At the same time,this paper combines the actual business situation of QD company,in the implementation process of online advertising fraud detection system,according to the overall program design,system workflow,main implementation steps to break down the task.The specific solutions and verification methods used in data collection,feature extraction,Sample processing,model training,etc.are described in the data level and algorithm level.This paper has certain versatility and operability for the research and method of online advertising fraud detection.It provides a practical reference for other enterprises in the industry to solve such problems and has certain reference significance.