This paper describes the results of an experiment designed to understand task-directed human explorative behaviour in a large music collection. The subject's task was to compile a music programme preferred in a specific context-of-use, e.g., romantic evening, party. Experimental conditions were defined in which subjects were provided with no music recommendations, randomly drawn recommendations, or algorithmically determined recommendations while carrying out the task. The provision of recommendations meant to improve performance in the compilation task. When recommendations were provided, subjects systematically selected, played back, and compiled fewer items by themselves, but instead made use of the recommendations. This observation was not coupled with a reduction in the amount of time spent on the compilation task. But when asked for their preference, subjects chose the provision of algorithmically determined recommendations above the provision of randomly drawn recommendations or no recommendations.
|Number of pages||10|
|Journal||IPO Annual Progress Report|
|Publication status||Published - 1997|