Building Privacy-Preserving Search Engine Query Logs for Data Monetization

Web Search Engines (WSEs) keep the record of the queries sent by their users in files named query logs. This kind of data is mainly used to improve the user experience by means of personalizing search results, auto-completing search terms, correcting spelling mistakes, etc. On the other hand, the companies that offer WSE services can also sell this data to obtain economic benefits, this is known as data monetization. Nevertheless, a lot of sensitive data from individuals can be inferred from their past queries (e.g. religion views, sexual preferences, full name, etc.), therefore, the uncontrolled disclosure of query logs may represent a serious privacy risk to the individuals who have participated in their generation. In order to prevent such a dangerous situation, query logs must be anonymized to break the link between any piece of sensitive data, its legitimate owner. This paper proposes a system that allows WSEs to compile privacy-preserving query logs that may be monetized. Well known WSEs such as Google receive, in average, 40000 queries per second, therefore, the proposed scheme has been designed to process queries as fast as possible, in a way that the privacy of the individuals is preserved but the data utility of the anonymized query logs remains maximized.