Context-aware music recommender systems: workshop keynote abstract

Recommender Systems (RSs) are software tools and techniques providing suggestions for items to be of use to a user [20]. In the music domain recommender systems can support information search and discovery tasks by helping the user to find relevant music items, for instance, new music tracks, or artists that the user may not even know [18, 9]. Several techniques have been proposed but most of the available systems use either contentor collaborativeor social-based approaches, or even more often, a hybrid combination of these three basic approaches [9, 7]. The recommendation algorithm is content-based, when the features of the music tracks that are liked by the user are considered when the system predicts what else the target user may like. Music features can be extracted directly from the music content, with signal processing techniques, or can be based on metadata (e.g., genre, year, author). Conversely, in collaborative-based approaches the system ignores the items’ descriptions, i.e., their features. It tries to find users with music preferences that are similar to those of the target user. Two users are estimated as similar by observing only the cooccurrences of the items in the sets of items liked/purchased by the two users. Then, the system recommends to the target user items liked by these similar users and novel to the target user. A third approach, which is called social-based, is emerging in the music domain. It is based on computing similarities among the items to be recommended (music songs or artists) through web mining techniques, or on exploiting social tagging information [8]. Social-based recommendations can be generated by using the similarities of artists that in turn can be computed using the social activity of the users, for instance by analyzing: the songs played by a community of users in the same listening sessions, or the tags assigned by users to songs or artists. The rationale of this approach is that items similar to those that the user liked will also probably be relevant to the user. However, notwithstanding the fact that music recommender systems are among the most common applications of recommendation techniques, there are very few music recommender systems that are capable to adapt their suggestions to contextual conditions important to predict the user’s preferences at a particular moment or situation. This is an important issue for a music recommender system, since people often seek music for a contextual situation like an occasion,