中文摘要：

在数据漏变得越来越严重的全球情形下面，我们在场瞄准数据漏预防(DLP ) 的一个基于可信赖的分发模型。在我们的模型，首先，分发者基于他的历史的行为计算用户可信赖；第二，根据用户可信赖和他的获得的文件集合，重叠漏了文件集合，分发者存取用户的概率是的故意的漏缝行为主观风险评价；第三，分发者作为一个客观元素评估用户站台危险；最后，分发者作决定是否基于综合风险评价散布文件。实验显示模型能区分不同类型的用户并且做比诚实用户要求的正在被否认的高被拒绝大部分的恶意的用户要求的概率，以便模型能够正当地阻止数据漏。

英文摘要：

Under the global circumstances where data leakage gets more and more severe, we present a trustworthiness-based distribution model that aims at data leakage prevention （DLP）. In our model, first, the distributor calculates the user＇s trustworthiness based on his historical behaviors; second, according to the user＇s trustworthiness and his obtained file set overlapping leaked file set, the distributor accesses the probability of the user＇s intentional leak behavior as the subjective risk assessment; third, the distributor evaluates the user＇s platform vulnerability as an objective element; last, the distributor makes decisions whether to distribute the file based on the integrated risk assessment. The experiments indicate that the model can distinguish users of different types and make the probability of malicious users＇ requirements being denied much higher than that of honest users＇ requirements being denied, so that the model is capable of preventing data leakage validly.