The Research Of The File Safety Monitor Based On The Feature Selection

With the development of the computer technology and the Internet, more and more information is stored and used as a form of the electronic document. Due to the openness of the electronic document, the file security incidents happened occasionally. Some of the documents contain sensitive information, which could be stolen and falsified. File security has become the one of the best important issues in computer security problems.File monitoring system could monitor the behavior of illegal tampering and falsified of the documents. The traditional monitoring systems are generally based on the keywords matching algorithm to search the files according to their names and the contents. But if the names and the contents are replaced or deleted, the files are not being able to found. In this paper, we proposed a local file searching technology based on the feature selection, which can apply to the monitoring system. The files are classified based on the feature selection, and its classification is unaltered if only the topic of the file is not changed. Thus, the searching system can still position the files. It improved the traditional method and made good advantages in file monitoring.In this paper, we first describe the overall framework and the workflow of the monitoring system, and then focus on the detailed design of the local file search module. We complete the module based on the classification technique of the feature selection. This article chose the algorithm of TF-IDF, and the improved algorithm is proposed due to the limitations of the algorithm itself. Advantage of improved algorithm is verified by experiments, and also verified the local file-based feature selection searching module which applied to the monitoring system can effectively protect the security of the file better.In the end, the article summarizes the research and indicates the future work.