中文摘要：

微博、微信等自媒体服务兴盛，危险预测成为微信息舆情管理的难题之一．基于SDN和MapReduce概念架构，结合虚拟蜜网技术，设计舆情倾向性检测模型；针对前端蜜罐机，设制舆情监测任务指令集，布局检测策略，完成分布式流量检测任务；通过虚拟嫌疑主题，针对大数据稀疏性困难，设计用户敏感行为特征集，实现微信息圈危害兴趣倾向的先验算法；最后对算法模型进行实践检验．实验表明，基于流量级和进程级关联的倾向性主题检测，检验效率较高，针对性强，能获得较好的监测效果，能为微信息舆情的主动性防范和舆情调节控制，提供重要的支持，所以，我们提出微信息进程与流量检测指令分布的倾向甘检测模型，以满足细粒度舆情监测与防御的需要．

英文摘要：

Since the media services such as microblog, WeChat flourish, risk prediction becomes the primary problem of micro- message public opinion management. This paper designs public opinion tendency detection model based on conceptual framework of SDN and MapReduce combining virtual honey net technology. For the front-end honeypot, this paper sets up public opinion monitoring task instruction set, and arranges detection strat- egy, then completes distributed traffic detection task.Through the virtual suspected theme, and in the face of the difficulties of large data sparse, it designs user sensitive behavior feature set, and realizes priori algorithm about interest tendency of micro-message circle.Finally,algorithm model is tested by practice.The practice proves that propensity subject detection based on flow level and process level association is efficient, and pertinence is strong ,and it can obtain good monitoring results.It provides important support for the active prevention of micro- message public opinion and regulation control of public opinion.