On the Identity Match Rate in Web Browsing History

Nowadays, various companies utilize Web browsing history in their business, such as web marketing. Since Web browsing history can be viewed as personal data, the companies may be able to identify a specific individual as the marketing analysis proceeds. In this paper, we assess the privacy risk of Web browsing history from the viewpoint of the identity match rate. The rate roughly evaluates the percentage of the following event in Web browsing history; Given a user's Web access data in a period, exactly the same user has the most similar Web access data in another period among all users in a group. If the identity match rate is high, there are many users whose Web access data in two distinct periods are the closest to each other and linkable. Therefore, it can be said that the privacy risk of such Web browsing history is high. In our privacy risk assessment, we pay attention to browsers' bookmarks. Bookmarks may have a higher privacy risk because they tend to reflect user preferences. In order to give evidence of it, we evaluate the identity match rate with the knowledge of bookmarks. Additionally, we take into account the timestamp and the category of the website for the evaluation. The analyses show that the identity match rate becomes the highest when bookmarks are used, which indicates the high privacy risk.