In this work, we address the problem of contextual recommendations by exploiting the concept of subspace clustering. Specifically, we pre-partition users that have rated subsets of data items similarly into clusters and we associate a context situation with each cluster. The cluster context is defined as any internally stored information that can be used to characterize the cluster members per se. Then, given a query context, we identify the clusters with the most similar context, and we use their members for making suggestions in a collaborative filtering manner. © 2016, Copyright is with the authors.
|