A day in MobileHCI Lab
Recommendation "SUCKS!"
ContextPlayer
Learning contextual music preferences for situational recommendations
SIGGRAPH 13
Presenter: Mobile HCI Lab Andy
Abstract
Contextual information enhance music recommendation experience
- Related Work & Their solutions
- Goals and Design
- UI
- System and Context
- Experiments and UX evaluation
- Conclusion
- How it influence me
Related Work
1. Most of the work available in the literature assumes
2. The current system does not process low-level music features for the recommendations
Their solutions
1. Use only implicit feedback, no assumptions
2. Mobile sensors and conditions
Goals and Design
combining mobile sensing
& the personal nature of music listening activities
“ smart! ”
&
Usable interface
Design : the suspender
Design : the stretch
UI
Design tool : interactive fabrication
System and Context
Examples
Experiments & UX evaluation
Pros
10/10 very interested in the recommandation
04/10 exceed their expectation
Most users enjoyed the UI
Cons
06/10 is initially confused
09/10 is not used to buy songs online
Many users have difficulty understanding the situation
63% wish the project have faster recommendation
Details
Details
Conclusion
Evaluating music context by location, activities, environment, music metadata and add UX into player redesign.
We have also identified the need to associate songs ;
We focus on building a model for music contextual recommendation
How it influence me
activity API Dumb but widely useWork in background. Be productive!
Thank you!
Your suggestions will be appreciated.
Copy of deck
By txshon Tseng
