Privacy-Enhancing Context-aware Search

With the quick development of Internet technology, WWW network owns more and more Web site and information resources. In the face of the vast information, the information which can be retrieved by users become more abundant, however, sometimes it is difficult for search engine to send the results which satisfy users. User search is the beginning of all the search behavior. If the retrieval system can’t understand user’s real intention, even though the back of a series of model and algorithm design is enough good, it’s impossible to make user satisfied. So, In the condition of user high quality requirements, the only way to solve "sameness" is to understand user need exactly.User search is in a particular search context, though now some retrieval system can use a part of context information and improve the problem of getting the same results, sometimes, the search results for search engine using the other valid context ineffectively don’t make user accept. Therefore, we need to analyse the effect and weight which all kinds of context bring in. On the other hand, as an important resource, the improper using of context will cause user privacy expose. When system aware effective context, what we also need to do is to analyse which kind of information is user’s privacy and user don’t want to share these information with Web sites, so further ensure the safety of user privacy. In view of the above two aspects involved in some related problems, the paper puts forward privacy-enhancing context-aware search, and the main work is as follows:(1) Design questionnaire. Context the paper defined include User context, Search context, Page context. Using the power of network, we send questionnaire on some Web sites, find out the valid context, mine the value of context, and analyse the influence and weight of different context to users’ search behavior.(2) Construct privacy-enhancing user interest model. Before constructing user model, we firstly analyse the characteristics, influence of the privacy information, and describe related research of privacy-protection; Secondly, describe the way to get user retrieval history, the characteristics of user interest model and its representation method; Lastly, design privacy-protection user model. The biggest difference between this model and common user model is to add the privacy-protection function, every node not only includes the semantic information of the node, the node weight, but also the risk value of node. According to the risk value of the node, the system will determine whether to prune this node or not and how to prune.(3) Construct the framework of privacy-enhancing context-aware search. Put the above user model to this framework and design the related algorithm. The framework elaborates the working principle of this system.(4) Validate and evaluate the system. This paper employs a way of inviting user to try out retrieval system and rate search results to test and verify the validity of privacy-protection user model, and analyses the all different outcomes detailedly.