Quick Question: Interrupting Users for Microtasks with Reinforcement Learning

Human attention is a scarce resource in modern computing. A multitude of microtasks vie for user attention to crowdsource information, perform momentary assessments, personalize services, and execute actions with a single touch. A lot gets done when these tasks take up the invisible free moments of the day. However, an interruption at an inappropriate time degrades productivity and causes annoyance. Prior works have exploited contextual cues and behavioral data to identify interruptibility for microtasks with much success. With Quick Question, we explore use of reinforcement learning (RL) to schedule microtasks while minimizing user annoyance and compare its performance with supervised learning. We model the problem as a Markov decision process and use Advantage Actor Critic algorithm to identify interruptible moments based on context and history of user interactions. In our 5-week, 30-participant study, we compare the proposed RL algorithm against supervised learning methods. While the mean number of responses between both methods is commensurate, RL is more effective at avoiding dismissal of notifications and improves user experience over time.

[1]  Mirco Musolesi,et al.  InterruptMe: designing intelligent prompting mechanisms for pervasive applications , 2014, UbiComp.

[2]  Brian P. Bailey,et al.  If not now, when?: the effects of interruption at different moments within task execution , 2004, CHI.

[3]  Alec Radford,et al.  Proximal Policy Optimization Algorithms , 2017, ArXiv.

[4]  Benjamin M. Good,et al.  Microtask Crowdsourcing for Disease Mention Annotation in PubMed Abstracts , 2014, Pacific Symposium on Biocomputing.

[5]  Aniket Kittur,et al.  CrowdForge: crowdsourcing complex work , 2011, UIST.

[6]  Janne Lindqvist,et al.  How Busy Are You?: Predicting the Interruptibility Intensity of Mobile Users , 2017, CHI.

[7]  Mirco Musolesi,et al.  Designing content-driven intelligent notification mechanisms for mobile applications , 2015, UbiComp.

[8]  Adam Tauman Kalai,et al.  A Crowd of Your Own: Crowdsourcing for On-Demand Personalization , 2014, HCOMP.

[9]  Mirco Musolesi,et al.  Intelligent Notification Systems: A Survey of the State of the Art and Research Challenges , 2017, ArXiv.

[10]  Moushumi Sharmin,et al.  Assessing the availability of users to engage in just-in-time intervention in the natural environment , 2014, UbiComp.

[11]  André van der Hoek,et al.  Microtask programming: building software with a crowd , 2014, UIST.

[12]  Lawrence B. Holder,et al.  Thyme: Improving Smartphone Prompt Timing Through Activity Awareness , 2017, 2017 16th IEEE International Conference on Machine Learning and Applications (ICMLA).

[13]  Ed H. Chi,et al.  Crowdsourcing for Usability: Using Micro-Task Markets for Rapid, Remote, and Low-Cost User Measurements , 2007 .

[14]  Jakub W. Pachocki,et al.  Learning dexterous in-hand manipulation , 2018, Int. J. Robotics Res..

[15]  S. Shiffman,et al.  Ecological Momentary Assessment (Ema) in Behavioral Medicine , 1994 .

[16]  Brian P. Bailey,et al.  Oasis: A framework for linking notification delivery to the perceptual structure of goal-directed tasks , 2010, TCHI.

[17]  Li Fei-Fei,et al.  ImageNet: A large-scale hierarchical image database , 2009, CVPR.

[18]  Kristjan H. Greenewald,et al.  Action Centered Contextual Bandits , 2017, NIPS.

[19]  Jason Flinn,et al.  The Case for Operating System Management of User Attention , 2015, HotMobile.

[20]  Martin Pielot,et al.  An in-situ study of mobile phone notifications , 2014, MobileHCI '14.

[21]  Aditya Ponnada,et al.  Microinteraction Ecological Momentary Assessment Response Rates , 2017, Proc. ACM Interact. Mob. Wearable Ubiquitous Technol..

[22]  Anind K. Dey,et al.  Leveraging Human Routine Models to Detect and Generate Human Behaviors , 2017, CHI.

[23]  Tadashi Okoshi,et al.  Reducing users' perceived mental effort due to interruptive notifications in multi-device mobile environments , 2015, UbiComp.

[24]  Martin Pielot,et al.  When attention is not scarce - detecting boredom from mobile phone usage , 2015, UbiComp.

[25]  Mirco Musolesi,et al.  My Phone and Me: Understanding People's Receptivity to Mobile Notifications , 2016, CHI.

[26]  Ariel Miller,et al.  Internet Usage by Patients with Multiple Sclerosis: Implications to Participatory Medicine and Personalized Healthcare , 2010, Multiple sclerosis international.

[27]  Martin Pielot,et al.  Dismissed!: a detailed exploration of how mobile phone users handle push notifications , 2018, MobileHCI.

[28]  Nuno Silva,et al.  Generating Human-Computer Micro-task Workflows from Domain Ontologies , 2014, HCI.

[29]  Mirco Musolesi,et al.  PrefMiner: mining user's preferences for intelligent mobile notification management , 2016, UbiComp.

[30]  Michael S. Bernstein,et al.  Twitch crowdsourcing: crowd contributions in short bursts of time , 2014, CHI.

[31]  Steve Benford,et al.  Investigating episodes of mobile phone activity as indicators of opportune moments to deliver notifications , 2011, Mobile HCI.

[32]  Gaël Varoquaux,et al.  Scikit-learn: Machine Learning in Python , 2011, J. Mach. Learn. Res..

[33]  Jaime Teevan,et al.  Chain Reactions: The Impact of Order on Microtask Chains , 2016, CHI.

[34]  Alex Graves,et al.  Asynchronous Methods for Deep Reinforcement Learning , 2016, ICML.

[35]  Brian P. Bailey,et al.  On the need for attention-aware systems: Measuring effects of interruption on task performance, error rate, and affective state , 2006, Comput. Hum. Behav..

[36]  Mani B. Srivastava,et al.  Sentio: Driver-in-the-Loop Forward Collision Warning Using Multisample Reinforcement Learning , 2018, SenSys.

[37]  Demis Hassabis,et al.  Mastering the game of Go without human knowledge , 2017, Nature.

[38]  Shane Legg,et al.  Human-level control through deep reinforcement learning , 2015, Nature.

[39]  Michael S. Bernstein,et al.  Break It Down: A Comparison of Macro- and Microtasks , 2015, CHI.

[40]  Sergey Levine,et al.  Deep reinforcement learning for robotic manipulation with asynchronous off-policy updates , 2016, 2017 IEEE International Conference on Robotics and Automation (ICRA).

[41]  Susan R. Fussell,et al.  Intelligent Interruption Management using Electro Dermal Activity based Physiological Sensor for Collaborative Sensemaking , 2017, Proc. ACM Interact. Mob. Wearable Ubiquitous Technol..

[42]  Anind K. Dey,et al.  ProactiveTasks: the short of mobile device use sessions , 2014, MobileHCI '14.

[43]  Thomas L. Griffiths,et al.  Faster Teaching via POMDP Planning , 2016, Cogn. Sci..

[44]  David Silver,et al.  Concurrent Reinforcement Learning from Customer Interactions , 2013, ICML.