A comparative evaluation of strategies for compiling music programmes

Abstract

A music compilation strategy, named Personalized Automatic Track Selection (PATS), has been implemented which aims at automatically compiling music programmes that are preferred in a specific context-of-use. It is additionally assumed that subsequent PATS-compiled programmes adapt to the designated context-of-use. The paper reports the results of an experiment designed to evaluate the appreciation of PATS-compiled programmes. More concretely, tests with naive users assessed the appreciation of PATS-compiled programmes compared to randomly assembled programmes in two different contexts-of-use employing a within-subject paradigm. Appreciation was expressed by the measures precision and coverage. The results of the experiment demonstrate that the PATS-compiled programmes contained more preferred and more varied musical content than randomly assembled programmes in both contexts-of-use. In addition, PATS-compiled programmes appeared to contain more preferred content over trials.