Blog: Noise Could Mask Web Searchers' IDs

Noise Could Mask Web Searchers' IDs
New Scientist (03/07/09) Marks, Paul

Microsoft researchers say that adding noise to search engine records could protect Web users' identities, and that implementing such a technique would be a major step toward provable privacy. Records of Web searches are extremely useful to software engineers looking to improve search technology, and can provide valuable insight for scientists exploring digital search behaviors. However, attempts to make search data anonymous have been mostly unsuccessful. Microsoft researchers Krishnaram Kenthapadi, Nina Mishra, Alex Ntoulas, and Aleksandra Korolova say they have developed a safe way to release search data. The researchers propose publishing data associated only with the most popular queries, so that specific, rarely performed searches, such as for individual names or unique interests, cannot be used to identify people. The researchers also inserted noise into the data by adding digits to the data's figures. Korolova says that adding the noise gives the data provable privacy, and the amount of noise added defines the level of privacy that can be guaranteed. She says the added noise strikes a balance between guaranteeing privacy and providing useful data sets.

View Full Article