Use of formal computational models for designing intelligent mobile device interfaces

ABSTRACT

This research examines the use of formal computational models to design intelligent device interfaces able to predict the function or application a user will use and automatically invoke this function. Computational models considered are Markov Chains, Markov Decision Process, Hidden Markov Model, Fuzzy Logic, and Bayesian Networks. Descriptive statistics obtained examine patterns of use on a mobile device. Usage profiles are modeled using the various computational models and verified with historical nonlearning data. This computational model framework is then deployed in field/laboratory trials where predictions are made on what the next user operation will be. Probabilistic predictions made by Markov Chains and Markov Decision Process are compared with context classification predictions made by Hidden Markov Model, Fuzzy Logic and Bayesian Networks. During the field/laboratory trials, each instance of use is a learning trial and the probabilities are recalculated for the models. In the next experimental phase, model predictions result in the automated delivery of certain functions or user operations. The benefits of automation are assessed in terms of task performance data (ease, accuracy, and speed of task completion) and user perceived usability using a questionnaire. The costs off automation are also assessed in terms of the costs of choosing a different operation and reversing the automated operation. In the last experimental phase, the ability to "undo" an automated operation is given and delivery of the next most highly probably operation is given as well as the capacity to "undo" each time or to come out of the automated prompting altogether. The benefits of this type of automation are assessed in terms of user performance and perceived usability. Future directions for research discuss how formal models can be used to design intelligent, highly-automated device interfaces and how best to design the automation to work in the users best interests.

REFERENCES

• 1: Dargie, W. A Distributed Architecture for Computing Context in Mobile Devices Department of Computer Science, Dresden University of Technology, Dresden, Germany, 2006.
• 2: Dix, A. Formal Methods: An introduction to and overview of the use of formal methods in HCI. in Perspectives on HCI: Diverse Approaches, 1995.
• 3: Eric Horvitz, Principles of mixed-initiative user interfaces, Proceedings of the SIGCHI conference on Human factors in computing systems: the CHI is the limit, p.159-166, May 15-20, 1999, Pittsburgh, Pennsylvania, United States [doi>10.1145/302979.303030]
• 4: Howard, R. A. Dynamic Programming and Markov Processes. The MIT Press, Cambridge, MA, 1960.
• 5: Rene Mayrhofer, An architecture for context prediction, Trauner, 2004
• 6: Stewart, W. J. Introduction to the Numerical Solution of Markov Chains. Princeton University Press, Princeton, NJ, 1994.
• 7: Harold Thimbleby , Paul Cairns , Matt Jones, Usability analysis with Markov models, ACM Transactions on Computer-Human Interaction (TOCHI), v.8 n.2, p.99-132, June 2001 [doi>10.1145/376929.376941]