A Privacy-Protecting Business-Analytics Service for On-Line Transactions

Abstract

Analysis of consumer-related and consumer-generated data is a very important way to measure the success of on-line retailing. The software packages for data analysis have two major shortcomings: (1) solutions are not offered as a service reachable by standard procedures over the Internet, but as isolated standalone applications or ERP system modules; (2) privacy restrictions need to be integrated into a framework of business analytics for Web retailers. The first aspect can be addressed with standardized developer software for Web services, but the second must consider privacy legislation, privacy specifications on Web sites (P3P), and data reidentification problems. These shortcomings are addressed by a proposed formal model of these problems and an implementation of the model as a declarative specification of privacy constraints, expressed as an extension of P3P. The constraints are complemented by a logic identifying the elements in a given set of Web analytics that might lead to data reidentification and therefore violate implicit privacy constraints. A Web-based service is presented that uses these components to automatically adapt the set of available Web analytics to an on-line retailer's P3P policy. The system was tested on a large data set from a major European multichannel retailer.