CoDeIn: A knowledge-based framework for the description and evaluation of reality-based interaction.

摘要

This thesis presents Cognitive Description and Evaluation of Interaction (CoDeIn), a framework that allows one to describe, compare, and estimate completion times for tasks that are designed in a variety of interaction styles. Task time estimates based upon evaluations performed with CoDeIn are consistently more accurate than evaluations carried out using existing model-based evaluation methods. This accuracy arises from several sources, including separating and distinguishing between different forms of knowledge necessary for performing the task. This distinction allows---as not provided by other methods---the ability to model non-expert task performance by estimating knowledge and its effect upon task completion time.; The accuracy of the CoDeIn methodology is supported by several experiments that compare the predictions produced by CoDeIn to actual measurements and predictions from the GOMSL methodology. To utilize GOMSL, several sub-models must be created, including a model of grasping and a model for iterative shuffling of physical objects.; I conclude that CoDeIn is in these circumstances more accurate and more expressive than GOMSL, due to the ability to account for and manipulate representations of knowledge.