Abstract: Based on the traffic analysis on Cernet2, a new discovery mechanism of games service provider IP address is proposed, including Boilerpipe text extraction algorithm and Stanford Chinese NLP, realize this new general method and computing architecture of college student internet addiction. Three kinds of Internet addiction behaviors-online games, video viewing and social networking sites visit, including the total hours spend, length of continuous addiction, frequency, selected time of addiction are analyzed. Concept of Internet addiction disorder index is proposed with quantization on the internet addiction disorder made through AHP. Finally, Internet addiction prevention system is designed and implemented.