文章基本信息

标题：Search Profiles Based on User to Cluster Similarity
本地全文：下载
作者：Zita Bošnjak ; Saša Bošnjak
期刊名称：International Scientific Journal of Management Information Systems
印刷版ISSN：1452-774X
出版年度：2009
卷号：4
期号：1
页码：003-007
出版社：University of Novi Sad
摘要：Privacy of web users' query search logs has, since the AOL dataset release few years ago, been treated as one of the central issues concerning privacy on the Internet. Therefore, the question of privacy preservation has also raised a lot of attention in different communities surrounding the search engines. Usage of clustering methods for providing low level contextual search while retaining high privacy-utility tradeoff, is examined in this paper. By using only the user`s cluster membership the search query terms could be no longer retained thus providing less privacy concerns both for the users and companies. The paper brings lightweight framework for combining query words, user similarities and clustering in order to provide a meaningful way of mining user searches while protecting their privacy. This differs from previous attempts for privacy preserving in the attempt to anonymize the queries instead of the users.
关键词：Search queries; clustering; privacy