In online content distribution services, users’ download history has been a primary data source for analyzing user interests, which enables the service providers to deliver personalized recommendations. Also, recent work in recommendation systems has shown that user interests are indeed time varying, and accurate profiling of user interest drifts can greatly improve the quality of recommendations. We present a visualization approach to analyzing user interest drifts from the download history, taking music as an example of downloadable media content, and study how to depict the underlying relevances among thedownloaded items to show the drifts.

We presented three new plots to visualize the music download history of one user, namely Bean plot, Transitional Pie plot, and Instrument plot. Bean plot displays the download history as colored “beans” showing music tracks with different genres, and “pods” standing for online sessions. Transitional Pie plot arranges the tracks by genre, and depicts the transitions between genres outside the “pie” and the collaborative relevances between tracks inside the “pie.” Instrument plot shows the tracks in chronological order as well as their collaborative relevances, and also displays the genre and release year of each track, together with their statistics. With these plots, one can analyze the user’s interest drifts more intuitively and efficiently.

Publications on the project:
Jingxian Zhang and Dong Liu. "Visualization of user interests in online music services." IEEE International Conference on Multimedia and Expo Workshops (ICMEW), 2014. pdf
Dong Liu and Jingxian Zhang. "Visual analyses of music download history: User studies." 2016 International Conference on Multimedia Modeling (MMM). pdf