Papers

WORKSHOP (INTERNATIONAL)

People's Interruptibility in-the-wild: Analysis of Breakpoint Detection Model in a Large-Scale Study

Kota Tsubouchi, Tadashi Okoshi(Keio University)

UbitTention 2017, 2017/9

Category:

Machine Learning Data Science HCI

Abstract:
In the advancing ubiquitous computing where users have been having an increasing amount of push-style information provision from lots of intelligent proactive services, detecting the users' current attentional status and/or interruptibility has been a significant issue. In our previous study[7] on interruptibility detection in "Yahoo! JAPAN" real world product for 21 days with more than 680,000 users, the result showed significant lower user response time in exchange for notification delivery delay for about 4 minutes on average. In this paper, we further analyze the features in the interruptibility detection model trained and updated during this study in nightly basis for 21 days and report the results.
Download:

People's Interruptibility in-the-wild: Analysis of Breakpoint Detection Model in a Large-Scale Study(External Site Link)