Research On P2P-based Collaborative Spam Sending Behavior Identifying Technology

Recently, most of the spam filters function on a single computer, so they can only obtain limited information and the efficiency is quite lower. Furthermore, many of spam makers apply constantly the new technology and approaches to make and send spam, which is challenging the traditional technique of anti-spam based on the email content. In order to solve this problem, this paper studies PSBIT (P2P-based collaborative spam sending behavior identifying technology).This paper studies the platform of identifying spam sending behavior for the purpose to scientifically design and implement the PSBIT. It analyzes the advantages and disadvantages of various resource searching patterns of the present P2P network, and designs Anti-Spam P2P network structure on the basis of super-peer by applying the mixed P2P network structure to the technology of spam filtering. It studies the query strategy of Anti-Spam P2P network and puts forward an improved mechanism of query and result returning, aiming at the problem of full connecting model in mixed P2P network structure bringing a great amount of redundancy searching message as well as the longer time of the returning of the searching results. The experiment manifests that this improved mechanism can reduce half of the redundancy querying information at rendezvous peer and the burden of the load of rendezvous peer. It can also enhance the returning speed of the result querying. It designs a collaboration spam filter based on JXTA in this platform of identifying spam sending behavior. According to the character of spam sending together with the principle of mail filtering at MTA communication process, it studies the identifying algorithm of spam sending behavior and the collaborative searching algorithm of suspicious mail. This paper defines the information table, the message and the advertisements in order to implement PSBIT. Also it describes the constructing process of anti-spam P2P network. In consideration of the disadvantages of manually setting peer status, the method of automatically confirming peer status is implemented on the basis of the capability of the email server. It describes in detail the process of implementing the algorithm of identifying the behavior of spam sending based on the P2P collaboration, and analyzes the cost of this algorithm theoretically.Finally, the PSBIT is experimented, and is compared with Foxmail based on bayesian in the spam filtering capability. The experiment result shows the PSBIT is just for the character of sending mails together without depending on the email content, the language type and the format analysis. This PSBIT can be also applied to filtering a large number of spams in the process of MTA communication, speeding up the process, improving the accuracy and saving a lot of systematical resources. It is much better than Foxmail in the recall, precision, F1, accuracy of filtering spam.