Paper Reading #14: A Framework for Robust and Flexible Handling of Inputs with Uncertainty

Commentary

See what I have to say about ___'s and ___'s work.

References

Schwarz, J., et al. (2010). A framework for robust and flexible handling of inputs with uncertainty. Proceeding of the Acm conference on user interface software and technology. New York: http://www.acm.org/uist/uist2010/.

Article Summary

This paper details a system that handles uncertain or ambiguous input. The researchers have devised a system that extends the conventional input system. Whereas in a conventional system, a user action either causes or does not cause a system action, in the uncertain system, all possible actions are taken into account and the most probable action is chosen. Actions do not cause final, irreversible changes to the system until temporary feedback is given to ascertain the intended input or the inferred action crosses a certain probability threshold. In the event that an action does not cross the "most probable" threshold, the user is given feedback of certain types when performing an action and may alter the action to generate the desired response. One possible temporary feedback type is detailed in the images below. As the user tries to select one slider, both are accidentally activated. A conventional system would just select one slider regardless of user intention, or do nothing at all. The uncertain system selects both sliders, gives temporary feedback on the possible state, and then allows the user to correct their input before a finalized action is taken.

Image courtesy of the above-cited article.

This system can handle uncertainty with both graphical and text input, including activation of multiple tiny buttons; inexactly placed input for scrolling, resizing windows, and dragging and dropping icons; multiple interpretations for spoken input to text translation; and greater ease of use for people with motor impairments.

Discussion

As with most projects that are at least latently philanthropic in nature, I really enjoyed this paper. For starters, gently correcting for erroneous input seems like a great idea, and it seems that this system does this without generating a large amount of overhead. We are already used to word processors automatically correcting our commonly misspelled words, or Google showing us results for what they think we really meant to search. Second, their results show a high rate of success in increasing ease-of-use for motor-impaired individuals, which is awesome. Admittedly, automatic "corrections" or "suggestions" can sometimes be pretty annoying; from what I've read, this system seems to strike a good balance.