Publishing Set-Valued Data Against Realistic Adversaries
The authors consider the privacy protection problem in publishing set-valued data. Set-valued data is a collection of transactions, and each transaction consists of an arbitrary number of items. Examples include shop-ping transactions, movie ratings, and web query logs. On one hand, set-valued data have a wide range of applications in data mining research. For example, customer behavior analysis is made possible by mining the shopping transactions, and movie rental companies have been successful in recommending movies to subscribers based on subscriber preference analysis obtained from the movie rating data.