Research On Filtering Spam Based On Global IP Reputation System

With the rapid development of Internet, Email, as one of the most widely used and popular service, has become a ubiquitous and arguably essential means of communication. But now this means of communication suffers challenge from Email spam. Larger number of spam not only waste network resource and people's time but spread various harmful information and network virus. So anti-spam research has become a hotspot issue.In this paper, we first introduce the background of the Email spam and the severe harm for the society to indicate that anti-spam research has far-reaching meaning and value. Then we introduce the actuality of anti-spam research, including policy, economy and technology. Since this research is built on email spam filtering technologies, we introduce interrelated technology in detail.Base on existed research fruit, we develop a new solution—global IP Reputation System. The main principle is that evaluating the reputation of global IP addresses, we can regard the reputation value of E-mail's IP address to distinguish whether a Email is Email spam. Compared with traditional filtering technology, this approach has the strongpoint of convenience, agility and reliability.This paper adopts BP neural network technology to establish data model. By analyzing system demand we choose several IP address attributes that have the great value in distinguishing spam as the input of neural network model. Then using some existed experiment data to train the model, we finally get reliable network model to judge IP address reputation degree.Global IP Reputation System is not only for IP address reputation evaluation, but can extract the IP characteristics involved in spam in order to integrate other spam filtering technologies easily. So based on the neural network above model, we do some further research to extract some rules to distinguish whether one email is spam or not, offering some valuable information for further research.