中文摘要：

预防发布信息泄密是公开发布信息安全领域面临的一个富有挑战的问题。信息之间存在的推导关系是导致泄密的关键因素,本文将这种推导关系作为检测泄密信息的重要依据,提出了基于指纹和推导模型的泄密信息检测方案。首先,将需要保密的信息（即秘密信息）分为篇章、段落、句子和词;然后,采用现有的指纹提取技术对篇章、段落、句子进行处理;再通过定义词的层次关系和推导关系,利用推导模型解决由于词的推导而导致的泄密问题;最后,给出了泄密信息的检测算法,并对该算法进行了理论分析和实验验证。

英文摘要：

The prevention from the leak ofrelease information is a major challenge in the field ofpublic release information security. The derivation relation between the information is a critical factor that causes the leak. In this paper, taking the derivation relation as an important means of detecting leak, a leak detection scheme based on the fingerprint and derivation model is presented. First, the information to be confidential is divided into chapters, paragraphs, sentences and words. Then they are processed respectively. As to the chapters, paragraphs and sentences, the existing fingerprint extraction technique is used. As to the words, the derivation model is constructed by defining the hierarchical relationships and derivation ones. Finally, the leak detection algorithm is presented. Furthermore, theoretical analysis and experimental validation is done.