Personalization through mask marketing

Abstract

Customized marketing, so called 1-to-1 marketing, is often viewed as the panacea of e-commerce. User profiles, such as click streams logging every site the user accesses, are exploited to generate a profile of interests of the users. Marketing measures like advertisements in form of banners or product presentation are tailored according to these profiles. This kind of customizing suffers from insufficient information about the current interests of the users: the profiles are only a collection of past actions of the users. Additionally, sometimes personalization is seen an invasion of privacy and often unsolicited advertisements unrelated to the users' real interests are cited as an example of the nuisances of 1-to-1 marketing. This shows the dilemma that 1-to-1 marketing faces: The users feel they already offer too much information about themselves. Currently, users solve this dilemma by disabling all technical means that enable user profiling, e.g. by disabling cookies in their Web browsers, using anonymity services, and providing fake identities to Web sites. We propose an approach based on peer-to-peer networks to allow marketing to a specific target group while preserving the privacy of users, we call it mask-marketing. In this concept a user assumes the identity of a mask representing his or her intentions. E.g., a user buying a Christmas present will use the "Christmas shopping" identity, afterwards the same user may use an "engineering" mask for an information search relating to his work. The users are able to present their preferences and intentions without these being attributable to their true identity. The masks, which are technically realized by sharing cookies, are distributed and updated using a peer-to-peer networking approach. This permits use of a mask by several different users. All actions performed by different users under the identity of one mask contribute to the profile of this mask. Creation and distribution don't require a central service. This is an advantage since the sensitive profile data is not concentrated in a central entity in the network, which could be viewed by privacy sensitive users with suspicion.