Abstract
A good use interface is central to the success of most products. Our research is concerned with improving an interface by making it adaptive - changing over time as it learns more about the user. In this paper, we consider the task of modifying a UNIX shell to learn to predict the next command executed as one sample adaptive user interface. To this end, we have collected command histories (some extensive) from 77 people, and have calculated the predictive accuracy for each of five methods over this dataset. The algorithm with the highest performance produces an average online predictive accuracy of up to 38%.