Portal de Periódicos da CAPES

Personalization and privacy

2001; IEEE Computer Society; Volume: 5; Issue: 6 Linguagem: Inglês

10.1109/4236.968828

1941-0131

John Riedl,

Sexuality, Behavior, and Technology

Personalization has been a hot topic or nearly a decade now, and many new products and advanced algorithms have emerged in that time. Several companies now sell tools such as recommender systems, which take input about users and products and generate recommendations about which products the users will like best. At their best, recommenders can be wonderful tools for users, helping them sort through myriad items they could read, buy, or watch to select those few that are most valuable to them. The algorithms that power these systems have evolved dramatically, and the best can produce rapid recommendations over data sets of millions of users and hundreds of thousands of products. The other edge of the sword is that recommender systems provide perfect tools for marketers and others to invade users' privacy. After all, recommenders; seek to learn everything about our preferences, including what we like to read, what we like to buy, how much money we spend, and what influences us to spend it. How a recommender deals with privacy decides whether its-users view it,as a boon or a bane. If the recommender only uses this information to help us find items to purchase on a Web site, we will probably value the feature - it might even bring us back to shop there again. On the other hand, if the Web site sells our information to other companies, so they can more effectively bother us. with phone calls at dinner time, we'll probably feel our privacy has been invaded. Privacy is a critical issue for recommender systems. In the end, personalization is an important factor in developing effective Web sites because it creates a user experience that is both compelling and sticky. The experience is compelling because it helps users find exactly the information, products, and services they need. It is sticky because a personalized Web site trains itself over time to serve its users better, which makes those users less likely to go to a new site that they would have to train all over again.