Hybrid Retrieval Approaches to Geospatial Music Recommendation.
Sprache des Vortragstitels:
Annual International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2013)
Sprache des Tagungstitel:
Recent advances in music retrieval and recommendation al-
gorithms highlight the necessity to follow multimodal ap-
proaches in order to transcend limits imposed by methods
that solely use audio, web, or collaborative ltering data. In
this paper, we propose hybrid music recommendation algo-
rithms that combine information on the music content, the
music context, and the user context, in particular, integrat-
ing location-aware weighting of similarities. Using state-of-
the-art techniques to extract audio features and contextual
web features, and a novel standardized data set of music lis-
tening activities inferred from microblogs (MusicMicro), we
propose several multimodal retrieval functions.
The main contributions of this paper are (i) a systematic
evaluation of mixture coecients between state-of-the-art
audio features and web features, using the rst standard-
ized microblog data set of music listening events for retrieval
purposes and (ii) novel geospatial music recommendation
approaches using location information of microblog users,
and a comprehensive evaluation thereof.