In this paper, we roughly divided the development of SNS in China into four stages:prototype SNS stage re-presented by blog, SNS entertainment stage represented by RenRen, micro information SNS stage represented by Weibo and WeChat, vertical mainstream SNS stage representing the future direction of development.Then from the perspective of “six elements business model”, and by introducing six dimensions into the analytical framework, i.e.the paying party in platform strategy, subsidies party, the same edge network effect, cross boundary network effects, switching costs, the paper outlines the development of Chinese SNS business model and analyzes the possible influence factors by contrasting the position aiming to blog, RenRen, Sina Weibo and Wechat, the platform cover, business system, key resources and capabilities, profit model, cash flow and enterprise value.Finally, the future prospects of Chinese SNS were discussed.

返利网站在发展和壮大过程中，面临的最重要问题就是吸引和留住用户。其中，留住用户是指促使用户在网上购物过程中持续地使用返利网站。通过借鉴扩展的信息系统持续使用模型为理论框架，并与关系营销学相结合，构建了返利网站用户持续使用意愿模型。实证研究的结果证实感知有用性、用户满意度、使用习惯、主观规范以及转换成本等因素会显著地影响返利网站用户的持续使用意愿，其中使用习惯和主观规范的影响作用最大；研究同时发现信任和感知易用性等因素的影响作用并不显著。

The three universities,i.e.,Dalian University of Technology,Huazhong University of Science and Technology and Peking University in mainframe applications talent training project,closely cooperate with industry enterprise to train mainframes software engineering talents.Absorbing the advanced teaching idea CDIO,they explore a set of "1-1-1-5" mainframe application talents training mode.The mode has achieved good results,greatly easing the shortage of mainframe talents.

The contemporary IRC botnet detection methods are not suitable for botnet detection under infrequently command and control interactions.To detect small stealthy botnet,a botnet detection model based on sequential analysis is proposed,which is a complement to contemporary passive detection technologies.Several probe methods and detection algorithms are discussed considering response types of clients,and average round of detection is analyzed,only small portion of command and control interactions are observed to declare single or multiple IRC bot.The results show that botnet detection is completed in expected round under controlled false positive rate and false negative rate.

To decrease the complexity of Botnet characteristic extraction and improve the speed of classification,a Botnet detection method based on Email characteristic match,which relies on neither Email detailed contents nor traffic analysis is presented.Raw emails are abstracted and Email characteristics are generated.Hellinger distance is used to find the most match characteristic in Botnet Email characteristic repository,then the Botnet that send the spam is classified.Experimental results show that the proposed method gained good accuracy and high efficency if enough spam Emails are trained and Botnet Email characteristic repository is well generated.

The existing Botnet techniques and detection methods are usually confined to specific Botnet.To improve the confidentiality of Botnet,the authors proposed a dynamic Botnet model described with directed graph,which can accommodate various Botnets.Several dynamic attributes of the proposed model were analyzed,such as exposedness,resilience,sustainability in detail,and then a bot abandon policy was presented.The experimental results indicate that the proposed method can decrease the Botnet's detection ratio and improve sustainability and resilience effectively.