Research On The Classification Method Of Document File Security Level

In the era of big data,the value of data assets has been greatly highlighted,and people also pay more attention to information security protection.This paper begins with the classification of document file security level,which aims to assist the inspectors in data leakage prevention through analyzing content and structure of the document files.This paper is concerned with text classification technology,steganography and steganalysis technique,document file structure checking technique and so on.Specifically,the security classification of document files can be divided into two parts:security classification for file content and security inspection for file structure.The file content classification is essentially a kind of text classification technology,so this paper conducts an in-depth study of existing text classification techniques,and in order to solve the problem of semantic loss and inaccurate feature extraction in traditional text classification system,the keyword-based word embedding representation and improved composite feature extraction model based on deep belief network are introduced.A contrast experiment proves that the change has effectively improved the accuracy of text classification.In the research of security check technology for file structure,we analyze the structure of Office file,PDF file,image file and compressed package file.And then combining with the technical characteristics of steganography,we conceived the way of hiding data in various types of files and put forward checking strategies accordingly.At the end of this paper,based on the research results of file content and structure,a file security level classification system is designed.The system can classify the content and check the structure for various types of document files.It can be used to data leakage prevention in outgoing control.