Skill metrics on a genetic graph as a mechanism for driving adaptation in an operating system interface

Abstract

Although simplified interfaces offer a convenient means of access to complex facilities, they do not help users to master the true complexity of a system. An interface which can gradually familiarize users with the commands necessary to perform complex tasks, adapting with them to reflect their individual skill acquisition, provides an attractive means of learning difficult command sets. This paper describes an adaptive shell to the cryptic command set of the UNIX ∗ ∗ UNIX is a trademark of AT&T Bell Laboratories. operating system. Implementation of adaptivity requires mechanisms which can adequately, but unobtrusively, model the individual user's progress in attaining skill.The use of a genetic graph to represent states of skill acquisition is discussed, together with the application of techniques from a simple Keystroke Level Model to generate suitable performance prediction metrics against which to determine skill levels. Experience with an exemplar for the shell in a normal work environment is presented to show that the approach taken can succeed in producing a system which is both sensitive to individual user variability and practical to use.