Study On Fake Website Monitoring And Management Technology

The popularity of computer network technology pushes rapid development of e-commerce and the growth in number of internet users promotes online transactions, but it has also led to amount of security problems. Phishing is specialized crime for users’account information in recent years, which not only caused serious economic losses and negative social impact, but also disrupted the network order seriously. In order to avoid more losses of businesses and individuals, researchers proposed defense program to against phishing attacks gradually, anti-phishing has become a new hot spot in network security field. There are many angles to prevent phishing attack which can be divided into three categories:filtering phishing e-mail, filtering domain and using heuristic feature to detect visited pages. Because most of phishing attacks would establish a fake website to get the victims’information, many defense programs aim at websites detection. Due to the phishing e-mail detection technology is limited in scope and black-and-white lists technology has time lag, researchers gradually focus on heuristic detection to increase the accuracy. This paper extracts URL and content features from the webpage then uses self-learning algorithm to classify the pages and judges the fake websites finally.This paper analyzes the strengths and weaknesses of the domestic and international anti-phishing technology, and pointed out key role of the fake websites in the anti-phishing identify. Aiming at the new features of the current fake websites, this paper proposed a kind of identifying method. First, extract the URL and web content as sensitive features of web page, then using SVM learning algorithm to train classifier and judging follow-up sample site.This paper tested the accuracy of SVM algorithm to detect fake websites and designed fake website detection process, which includes URL information reading, Web page content reading, sensitive features extraction, features pretreatment and SVM classifier.The innovation of this paper:1. Improved methods of detecting fake websites. Using URL and content features of Web page to detect fake Websites can effectively improve the accuracy of detecting fake Websites. 2. Combining self-learning SVM classification method. SVM algorithm has very superior performance when classify small sample. Existing detecting methods apply multi-SVM algorithm to the Web image classification, here SVM algorithm is used to the page text feature content which get detection effects for non-image webpage.3. Using improved representation of the feature quantity. SVM sample input label expressed by Boolean value using, this paper use the formula calculate the characteristic amount of weight which reflects importance degree of each feature in the detection.