The Discovery And Protection Of User Privacy Infringement In Web Environment

With the popularity of web search,the flat tendency in accessing web resources enlarges the risk of privacy leak with its spread.Different from conventional data security technologies,privacy preserving applies to prevent released documents from occasionally carrying sensitive and user-forbidden information.Meanwhile,the challenge of this work is that there does not exist absolute safety in user privacy:To enjoy better web services(e.g.personalized web services),users have to trade parts of their personal information for the better understanding on each individual from web applications.It is the trade-off between better web services and safer user privacy.The contribution of this work is to find the leverage,which ensures those parts of traded personal information do not infer extra privacy infringement risk.The paper focuses on preserving web user privacy from varied points of view, refers up-to-date worldwide related work,and collects three aspects of the author's previous work,namely a) the detection of privacy leak;b) an approach to preserve privacy of user preference based on domain signatures;and c) the unified protection framework on web user privacy.The first work of privacy preserving is to detect privacy infringement.This paper proposes a novel method probing users' online motivation from their opened & de-sensitized navigation history.The method bases on the fact that more users get used to conducting their surfing behavior via web search.It induces stronger connectivity among the context of web pages browsing and finally causes privacy leak.The experiment later shows the relatively high precision in probing user surfing motivation,which proves the effectiveness of this method's in finding privacy leakage as early as possible.Secondly,as another kind of privacy,web user preference is also widely used in personalized web services.Preference includes large amounts of user subjective data, which usually infer rich user privacy.Currently,the protection on preference data only relies on server-sides self-conscious.This paper proposes a strategy de-identifying web users by domain signatures to realize privacy protection.The final section illustrates the privacy protection framework in web environrnent, preparing knowledge for developing applications under this research topic.The paper depicts two aspects of this unified framework.One is to detect whether the documents to be released contain latent sensitive information beforehand.The other is how users monitor the sensitive information after data are released to the web,and then ahead of malicious adversaries,stop the further propagation of sensitive data afterwards.