Intelligent interface design: An empirical assessment of knowledge presentation in expert systems

Abstract

This research evaluates intelligent interface requirements for knowledge presentation in an expert system used for diagnostic problem solving. In a field study, interactions between employee expertise, knowledge presentation format (procedural vs. declarative), question type (requiring abstract vs. concrete knowledge organization), and task uncertainty are examined for employee problem-solving and decision-making performance (speed and accuracy). Also evaluated are confidence in system recommendations and lines-of-reasoning, as well as user satisfaction with the system interface. The study provides findings that are discussed within the context of intelligent interface requirements for organizational information systems. The results show that high-skill users perform significantly faster and more accurately when solving the problems and have self-reported confidence ratings that are higher than those of low-skill users. The expert system, however, has a greater impact on improving performance for low-skill users than for high-skill users. A relationship is found between skill level and task uncertainty indicating that different skill-level users require different presentation formats paralleling their conceptual representations of the problem. The interaction between skill level and knowledge organization is confirmed with results showing that low-skill users perform faster than high-skill users on questions requiring concrete knowledge organization; whereas high-skill users perform better when presented with questions requiring abstract knowledge organization.