Development and validation of a conceptual model for a skill-based adaptive human-computer interface.

摘要

The objective of this study is to model and design a skill adaptive interface for human-computer interaction. The adaptation applies to the changing skill levels as users become familiar with an interface. Commands and menus, which require different human memory abilities (recall vs. recognition) and allow different operating speeds, are the most frequently used interface modes. A three-dimensional model is presented to predict the relations among interface style, users' skill levels at using an interface, and users' domain conceptual knowledge (CK). Based on this model, a skill adaptive interface, which is a dynamic combination of menus and commands, was designed and implemented. An experiment was conducted to test five hypotheses concerning the validity of the model and the usability of the adaptive mechanism. Eighty subjects participated in the experiment. The dependent variables were the time to complete the assigned tasks, the number of steps that were used to complete the tasks, the proportion of steps used in the menu mode, the perceived memory load, and the satisfaction ratings. The independent variables were the interface style (Adaptive, Menu Only, Command Only and static Hybrid), the user's skill level at using an interface, and the user's CK level of the operating systems. A between-subject design was used for the interface style. It was confirmed that (1) high CK users with low skills performed faster by using the Menu Only interface than the Command Only interface, (2) the Command Only interface was the best for high-skill users, (3) the Adaptive interface could integrate the advantages of both command and menu modes for the high CK users and (4) the Adaptive interface yielded significantly faster performance than the Hybrid interface for the high CK users. It was demonstrated that if the proposed adaptive interface mechanism was applied properly, users could improve their performance time significantly. However, the experimental results did not support the hypothesis that domain conceptual knowledge was a more significant contributor to user performance than the use of different interface styles.