Method for retrieving confidential information of file and judging and marking security classification based on content correlation

Abstract

A method for retrieving confidential information of a file and judging and marking the security classification based on content correlation is mainly characterized by comprising the following steps of: selecting corresponding confidential keywords and a matched rule base according to the range to which a document belongs; accurately and comprehensively searching confidential information in the document by carrying out word segmenting operation on the document contents and constructing a semantic correlation network; carrying out security classification judgment and encryption processing on each confidential information point according to a secret-level setting rule in a secret-level setting rule base; then giving out corresponding security classifications by integrating the security classifications of various confidential information points; and finally, marking related confidential information in the documents. According to the method, the confidential information in the document can be efficiently searched and the security classification of the document is accurately judged, so that the workload of manual secret-level setting is reduced, the efficiencies of secret-level setting work and the security classification appraisal work are increased, the random secret-level setting phenomenon in the secret-level setting process is effectively prevented and new confidential keywords and matching rules can be automatically learned, and further the precision of retrieval and security classification judgment is continuously improved.

Description

technical field [0001] The invention relates to a method for retrieving, judging and labeling secret-related content of documents based on content correlation for specific content in electronic documents. Background technique [0002] In the current secrecy work, the management and use of confidential documents is a very important link, and the determination of the confidentiality level of documents is a very important task. The current classification of confidentiality all depends on manual implementation, which makes the period of determination work too long and the accuracy is not high, and the determination of the classification of confidentiality involves the entire document, which brings some problems to the reference of some non-classified content in confidential documents. Great inconvenience. For this reason, it is particularly important to be able to rely on a computer to automatically and effectively judge whether the content of a file is classified, which specif...

AI Technical Summary

Problems solved by technology

[0003] 1. The accuracy is not high

Some keywords and feature points are confidential only in a specific context, and simple matching of keywords and feature points can only find out whether the word and feature point are contained in the file, and cannot effectively judge the location. The context of the text, therefore, many times the "classified content" found out is not actually classified

[0004] 2. Integrity is not strong

There are many ways to express the content of the document, including text, pictures, video, audio, etc. In terms of text alone, there are thousands of different ways to express a thing, which makes it impossible to use some fixed keywords or feature points to cover all related issues. Confidential information, keywords and feature points can only serve as an outline, and the method based on simple search and matching may miss a lot of content that does not contain keywords and feature points, but contains confidential information, which brings great harm to confidentiality work. to hidden danger

[0005] 3. The availability of search results is not strong

Using the existing method to find out the file, the user can only know whether the entire file is classified or not, and cannot specifically give which word, sentence, and paragraph in the file is classified or classified. What are the classification levels, which will cause a lot of inconvenience to quote the contents of this document in the future

[0006] 4. Does not have the function of automatic learning

However, the existing methods can only search on the basis of existing secret-related documents, and there is no way to involve new secret-related content

Images