Network Analysis And Filtering Technology Research Based On The Content

With the rapid development of the Internet, the network has become an important platform for people's learning, working and living. The network provides us great convenience. At the same time, there are very great negative effects, such as the spread of illegal information, leaking of secret information, or garbage e-mail. These negative effects have threatened the network's security and efficiency. It seems vital important to the network's further development to analysis the information and filter the illegal information, or, to preserve the proof of the leaking of secret information.The research of the network analysis and filtering technology based on content has been carried on in the thesis, in order to filter the network's information security. In the thesis, a general model of the network analysis and filtering system based on content has been suggested, and a project based on the general model has also been put forward and analyzed. As for the real-time data acquisition and preprocessing being suggested in the general model, some key technology, especially TCP restoring, have been studied and solved.There are some faults in the classic network analysis and filtering method based on the keywords, a new method based on SVMs is suggested in the thesis to deal with the faults. In the new method, the building of vector space for Chinese text content, the rapid vectorization for Chinese text content and some other problems should be solved. In the thesis , some data structures and algorithms are suggested to solve them.Considering the network's demand on real-time processing and the complexity of the content, it is a very difficult task to implement the network's analysis and filtering technology based on the content. There is only a small part task has been done in the thesis and many problems need to be studied and solved in the future. These problems include implementation SVMs in the system to solve some particular question, the classification and recognition of the graphic content based on the SVMs, etc.