Beta 1


Title People, Places and Playlists - Modeling Soundscapes in a Mobile Context
Author Zandi, Nima
Handler, Rasmus
Supervisor Larsen, Jakob Eg (Cognitive Systems, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
Petersen, Michael Kai (Cognitive Systems, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
Institution Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark
Thesis level Master's thesis
Year 2010
Abstract In this thesis we aim to examine whether music listening data combined with raw sensor data from a mobile device, can identify a mobile context and facilitate social interaction. Aiming to model not only the people, places and time that make up our mobile context, but also the constantly changing frame of mind reflected in the music we listen to, music has been utilized as yet another embedded sensor. A field study has been conducted with a handful of participants, constituting a small scale social network, carrying a smart phone for two weeks. All phone activities and data from embedded sensors was recorded, along the music being played on the device. In order to be capable of extracting contextual knowledge from the music being played, social tags retrieved from Last.fm has been utilized. Using a model that is capable of extracting the underlying topics behind the most frequently co-occurring words from the tag clouds, allows us to model the emerging concept of Soundscapes that captures the underlying semantic aspects of songs within a playlist. With visualization and similarity capabilities, the Soundscapes has not only been used to support the various scenarios related to music consumption, it has also provided the basis for identifying which conditions that determine music preferences in a given context. We report initial indications that listening patterns in terms of music genre preferences are influenced by whether the user is in a static environment or on the move. Applying a simple decision tree algorithm, in order to identify what contexts determine the preferences, indicates that our listening patterns change over time. This suggest that music applications utilizing context information must be designed to adapt to our shifting preferences as they continuously evolve. Furthermore, a conceptual model assembles the concepts in order to support music navigation, recommendation and exploration for the individual as well as in a social context. Thus, patterns emerge that indicate how it might be possible to share music by traversing the social graph and finding other users (friends) being in a similar context or listening to similar playlists. This leads us to a proof-of-concept application, SocialContextPlaylist, aiming to demonstrate how context-awareness, music similarity and social relations can be exploited in order to enhance the social experience of future mobile music applications by facilitating social interaction. The application is meant to serve as a foundation for further analysis and development of new creative innovative solutions in the emerging domain of future mobile entertainment applications and services.
Imprint Technical University of Denmark (DTU) : Kgs. Lyngby, Denmark
Series IMM-M.Sc.-2010-25
Fulltext
Original PDF ep10_25.pdf (12.95 MB)
Admin Creation date: 2010-05-11    Update date: 2010-11-25    Source: dtu    ID: 261940    Original MXD