Andreu Vall,
"Listener-Inspired Automated Music Playlist Generation"
: Proceedings of the 9th ACM Recommender Systems Conference (RecSys), 9-2015
Original Titel:
Listener-Inspired Automated Music Playlist Generation
Sprache des Titels:
Englisch
Original Buchtitel:
Proceedings of the 9th ACM Recommender Systems Conference (RecSys)
Original Kurzfassung:
The objective of this PhD research is to deepen the understanding
of how people listen to music and construct
playlists. We believe that further insights into such mechanisms
can lead to enhanced music recommendations. We
research on the exploitation of user-generated data in the
context of on-line music services, since it constitutes a rich
and increasing source of information of user behavior. The
research carried out so far has centered on the scenario of
producing a single artist recommendation. Concretely, in
this paper we show how to mitigate the cold-start problem
for new artists, elaborating on our findings on the combined
effect of users' listening histories and users' tagging activity.
As future research, we will investigate how improved tech-
niques to exploit user-generated data can also be applied
to the task of producing sequential recommendations, like
playlists. We are particulary interested in creating music
playlists similarly as users would do, and in finding mechanisms
to make such music streams adapt to users' feedback
on-line.