A hybrid social-acoustic recommendation system for popular music

  • Authors:
  • Justin Donaldson

  • Affiliations:
  • Indiana University, Bloomington, IN

  • Venue:
  • Proceedings of the 2007 ACM conference on Recommender systems
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

Recommendation systems leverage several types of information relating to a recommendable item. The recommendation methods are often based on the analysis of how a set of users associate or rate a given set of items, but they can also focus on the analysis of how the content of the items is related. This paper discusses a hybrid recommendation system for music - a system that leverages both spectral graph properties of an item-based collaborative filtering association network as well as acoustic features of the underlying music signal. Both features are balanced appropriately and used to disambiguate the music-seeking intentions of a user.