Sök:

Rekommendationssystem för interaktiva musiktjänster

en utredning av aktuella trender och attityder gentemot framtidens rekommendationssystem diskuterat ur forskar-, industri- & användarperspektiv


Parallel to new technological advancements, including the development of interactive recommender systems, digital music services on the Internet, such as Spotify, have in a short time span gradually replaced former physical media such as CDs and MP3-players and become pioneers on a market with an otherwise uncertain future. By subscribing to the service, users have access to a vast library of music and from any device with Internet access. In order to provide a satisfactory and useful musical experience, different ways of coping with the vastness is required - which is precisely what recommender systems are for. Improved filtering techniques and use of devices built in sensors, make way for new opportunities of more precise and contextually tailored suggestions for track selection, which in turn leads to potential privacy issues when using interactive music services. But where do we stand today? By compiling, exemplifying and discussing three different perspectives with relating to the development of interactive music recommender systems, the objective for this paper is to provide a for science, industry and user, objectively nuanced account of current potentials and expectations. A conducted survey among Swedish-speaking Spotify users as well as a performed asynchronous interview with an industry expert at Spotify, make up the foundation for an analysis based on dimensions and qualities of satisfaction related to the evaluation process of recommender systems. The study shows that users' are somewhat cautious toward future technological implementations, which may potentially infringe on their privacy, yet still accepting, due to the potential of a more personalized music experience. From an industry perspective, the possibility of offering contextualized recommendations is considered potentially valuable. Thus, conditions for realizing a more accomplished recommender system exists, but with some reservations - primarily requests for increased transparency regarding what information recommendations are based upon, and the ability for users to self-regulate what information is made available to the service - which in turn the recommendations shall be based on. 

Författare

Magnus Johnson Rasmus Svensson

Lärosäte och institution

Södertörns högskola/Institutionen för naturvetenskap, miljö och teknik

Nivå:

"Kandidatuppsats". Självständigt arbete (examensarbete ) om minst 15 högskolepoäng utfört för att erhålla kandidatexamen.

Läs mer..