October 2020
Volume 20, Issue 11
Open Access
Vision Sciences Society Annual Meeting Abstract  |   October 2020
Why are you reading this? Predicting reading goal and familiarity from people’s mobile interaction behaviors
Author Affiliations & Notes
  • Sungjin Nam
    University of Michigan
  • Zoya Bylinskii
    Adobe Research
  • Christopher Tensmeyer
    Adobe Research
  • Rajiv Jain
    Adobe Research
  • Curtis Wigington
    Adobe Research
  • Tong Sun
    Adobe Research
  • Footnotes
    Acknowledgements  This work was funded as a part of an internship work at Adobe Research
Journal of Vision October 2020, Vol.20, 951. doi:https://doi.org/10.1167/jov.20.11.951
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Sungjin Nam, Zoya Bylinskii, Christopher Tensmeyer, Rajiv Jain, Curtis Wigington, Tong Sun; Why are you reading this? Predicting reading goal and familiarity from people’s mobile interaction behaviors. Journal of Vision 2020;20(11):951. doi: https://doi.org/10.1167/jov.20.11.951.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

People read for different reasons and in different contexts. The reading behavior of a paralegal skimming a contract for specific legal phrases will look quite different than that of a student learning about unfamiliar concepts from a textbook. While the reading of printed material is still commonplace, reading in a mobile context is becoming increasingly popular. From an experimental perspective, mobile devices provide an effective platform for capturing behavioral data outside the laboratory, non-invasively and at scale. In this work, we study reading on mobile devices while measuring readers’ attention and behavior without the use of equipment like eye trackers. Specifically, we measure the effects of reading goal and familiarity on the mobile interaction behaviors (touch, scroll, reading time) of 285 crowdsourced (Mechanical Turk) participants reading approximately 500-word articles on mobile devices. We manipulated the familiarity condition by exposing participants to article summaries during a training phase, either related or unrelated to articles that they read during the main study. We manipulated the reading goal by changing instructions participants received, to encourage either ‘literal’ or ‘contextual’ reading. Our findings suggest that features based on touch locations can be used to distinguish between familiarity conditions, while scroll-based features and reading time can be used to differentiate reading goal conditions. We also built statistical models that, for a given participant, predict the familiarity level (55.6%) and reading goal (70.2%) with greater accuracy than a baseline model (51.1% (n.s.) and 51.7% (p<0.001), respectively). Moreover, individual differences in scrolling behaviors on a mobile device during reading affect the prediction of reading goals when using scrolling speed and idle time as features. The results of our studies motivate promising future investigations of reading on mobile, towards building predictive models of individual readers, as building blocks for customized reading experiences.

×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×