基于数据挖掘和web信任模型的属性认证系统设计实现
发布时间：2018-10-15 12:06
【摘要】：社交网络服务(Social Network Service,SNS)以为用户提供数据为基础,提供了多种多样的社交网络服务方式,但反之社交网站中虚假信息也越来越成为网络中的安全隐患。社交网络中用户自身可以随意修改自己的属性信息,造成了属性信息的不可信,社交网络中的用户属性信息危机越来越严重。现如今的属性权威系统中利用CA认证的方式,为每一个用户颁发属性证书,经过CA认证的属性证书都是有效而且可信的,但是这种情况只适用于小型封闭型系统,对于大型的社交网络来说这种情况并不适用。基于以上背景,本文基于国内最大的微博平台,针对用户属性认证问题,分析用户的社交网络行为,对多个领域的研究理论和研究成果都进行了调研,研究和比较分类。针对社交网络用户属性可信度低且缺少评价这一问题,通过借鉴标准的PGP利用web of trust来验证公钥的有效性的思想,提出了基于PGP的WEB信任模型来给用户提供属性认证,给用户的属性信息提供了安全的属性认证途径。因为基于社交约束和WEB信任的方法需要人的参与,当服务刚上线用户不多时,PGP WEB信任模型难以在稀疏网络确保对任一请求找到一条信任链,所以我们通过对用户的社交行为进行数据挖掘,预测用户属性进而弥补PGP WEB信任模型在模型初期的问题,通过社交网络用户属性信息挖掘的结果帮助提高WEB信任网的密度,并检测和提示恶意评价。在设计用户属性信息挖掘模型的时候,本文充分考虑到了用户的属性信息的相关性,引入了多标签分类对用户进行属性预测,对性别预测的准确率超过了 80%,年龄和职业预测的准确率也超过了 70%,多标签模型不仅仅优化了数据挖掘进行属性预测的时间性能,数据挖掘模型的可抽象性和可扩展性都得到了有效地提高。通过结合社交网络用户属性信息挖掘和基于PGP的WEB信任模型两种方式来给用户提供属性认证可以满足用户属性认证的安全性和准确性的需求。
[Abstract]:Social network service (Social Network Service,SNS) provides a variety of social network services based on providing users with data, but on the contrary, false information in social networking sites is becoming a security hazard in the network. In social network, users can modify their own attribute information at will, which leads to the disbelief of attribute information, and the crisis of user attribute information in social network is becoming more and more serious. In today's attribute authority system, attribute certificates are issued to every user by means of CA authentication. The CA certified attribute certificates are valid and credible, but this is only true for small closed systems. This is not the case for large social networks. Based on the above background, based on the largest Weibo platform in China, this paper analyzes the social network behavior of users for the problem of user attribute authentication, and makes a research, research and comparative classification on the research theory and research results in many fields. In order to solve the problem of low reliability and lack of evaluation of user attributes in social networks, a WEB trust model based on PGP is proposed to provide attribute authentication to users by referring to the idea that standard PGP uses web of trust to verify the validity of public key. It provides a safe way to authenticate the user's attribute information. Because the approach based on social constraints and WEB trust requires the participation of people, it is difficult for the, PGP WEB trust model to find a trust chain for any request in sparse network when the service is only a few online users. So we use the data mining of user's social behavior to predict the user's attribute and then make up the problem of PGP WEB trust model in the early stage of the model. We help to improve the density of WEB trust network through the result of the user's attribute information mining of social network. And detect and prompt malicious evaluation. When designing the user attribute information mining model, this paper considers the correlation of the user's attribute information, and introduces multi-label classification to predict the user's attribute. The accuracy of gender prediction exceeds 80%, and the accuracy of age and career prediction exceeds 70%. The multi-label model not only optimizes the time performance of data mining for attribute prediction, The abstractness and extensibility of data mining model are improved effectively. Through the combination of social network user attribute information mining and WEB trust model based on PGP to provide attribute authentication to users can meet the needs of security and accuracy of user attribute authentication.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP393.0;TP311.13
本文编号：2272504
[Abstract]:Social network service (Social Network Service,SNS) provides a variety of social network services based on providing users with data, but on the contrary, false information in social networking sites is becoming a security hazard in the network. In social network, users can modify their own attribute information at will, which leads to the disbelief of attribute information, and the crisis of user attribute information in social network is becoming more and more serious. In today's attribute authority system, attribute certificates are issued to every user by means of CA authentication. The CA certified attribute certificates are valid and credible, but this is only true for small closed systems. This is not the case for large social networks. Based on the above background, based on the largest Weibo platform in China, this paper analyzes the social network behavior of users for the problem of user attribute authentication, and makes a research, research and comparative classification on the research theory and research results in many fields. In order to solve the problem of low reliability and lack of evaluation of user attributes in social networks, a WEB trust model based on PGP is proposed to provide attribute authentication to users by referring to the idea that standard PGP uses web of trust to verify the validity of public key. It provides a safe way to authenticate the user's attribute information. Because the approach based on social constraints and WEB trust requires the participation of people, it is difficult for the, PGP WEB trust model to find a trust chain for any request in sparse network when the service is only a few online users. So we use the data mining of user's social behavior to predict the user's attribute and then make up the problem of PGP WEB trust model in the early stage of the model. We help to improve the density of WEB trust network through the result of the user's attribute information mining of social network. And detect and prompt malicious evaluation. When designing the user attribute information mining model, this paper considers the correlation of the user's attribute information, and introduces multi-label classification to predict the user's attribute. The accuracy of gender prediction exceeds 80%, and the accuracy of age and career prediction exceeds 70%. The multi-label model not only optimizes the time performance of data mining for attribute prediction, The abstractness and extensibility of data mining model are improved effectively. Through the combination of social network user attribute information mining and WEB trust model based on PGP to provide attribute authentication to users can meet the needs of security and accuracy of user attribute authentication.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP393.0;TP311.13
【参考文献】
相关硕士学位论文 前2条
1 吴伊萍;中文微博情感分类研究[D];华侨大学;2013年
2 张智;数据挖掘在高校学生综合测评体系中的研究与应用[D];江西农业大学;2011年
，本文编号：2272504
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2272504.html
