Abstract
Assistance systems are beginning to become necessary components in human-computer systems. The present research describes an initial attempt at providing a performance-based assistant for a complicated information retrieval task. The assistant was implemented based upon an expert profile methodology. The resultant expert performance model was used to diagnose inappropriate novice search strategies. A suggestive and forceful assistant were compared against an unassisted novice subject group. The suggestive assistant highlighted appropriate search procedures for novice subjects on a keypad with a touch entry display. The forceful assistant also highlighted appropriate search procedures but eliminated inappropriate search procedures from the touch keypad. The experiment demonstrated a marked improvement in the search strategies and performance of assisted novices. These enhanced search styles were not without a cost. In this implementation, both the suggestive and forceful assistants were very intrusive on trials with advice. Consequently, further research should address the development of assistance systems that naturally transfer expertise to inexperienced users.
Get full access to this article
View all access options for this article.