Optimization of switch virtual keyboard by using computational modelling
Abstract
In this thesis, I first reviewed some keyboard technologies used by people with motor difficulties, and described design elements that influence efficiency. I cast the design of a switch keyboard as an optimization problem, and arrangement of keys on such a keyboard as a Mixed Integer Programming problem. One significant variable in the MIP problem, the error rate, is related to several other variables. I treated modeling of the error rate as a parameter estimation problem, and used a data mining method. I designed HCI experiments to gather data for parameter estimation, using Bayesian logistic regression model. The empirical data and error rate modeling allowed for construction of several different types of keyboards. These different keyboards were compared and evaluated with regard to their use by people with motor difficulties.
Degree
M.S.
Advisors
Francis, Purdue University.
Subject Area
Cognitive psychology|Computer science
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.