The set of users has first to be partitioned into a number of groups equal to the number of recommendations. Since in our application scenario groups do not exist, unsupervised classification (clustering) is necessary. Users are clustered considering the ratings expressed for the evaluated items. It was recently highlighted in [14] that the k-means clustering algorithm [12] is by far the most used clustering algorithm in recommender systems. This task detects groups by clustering users with the k-means clustering algorithm. The output of the task is a partitioning of the users in groups (clusters), such that users with similar ratings for the same items are in the same group and can receive the same recommendations.