SecWeb privacy-preserving web browsing monitoring with w-event differential privacy

Abstract

Nowadays aggregated web browsing histories of individual users have been collected and extensively used by Internet service providers as well as third-party researchers, due to their great value to data mining for in-depth understanding of important phenomena, such as suspicious behavior detection. While providing tremendous benefits, the release of private users’ data to the public will pose a considerable threat to users’ privacy. Sharing web browsing data with privacy preservation has so far received very limited research attention. In this paper, we investigate the problem of real-time privacy-preserving web browsing monitoring, and propose SecWeb, an online aggregated web browsing behavior monitoring scheme over infinite time with theoretical privacy guarantee. Specifically, we propose an adaptive sampling mechanism and an adaptive budget allocation mechanism to better allocate appropriate privacy budget to sampling points within any successive w time stamps. In addition, we propose a dynamic grouping mechanism that groups web pages with small visits together and adds Laplace noise to each group instead of single web page to eliminate the effects of perturbation error for the web pages. We prove that SecWeb satisfies w-event differential privacy and the experimental results on a real-world dataset show that SecWeb outperforms the state-of-the-art approaches.