A performance profile methodology for implementing assistance and instruction in computer-based tasks

✍ Scribed by Jay Elkerton; Robert C. Williges

• Publisher: Elsevier Science
• Year: 1985
• Weight: 985 KB
• Volume: 23
• Category: Article
• ISSN: 0020-7373
• DOI: 10.1016/s0020-7373(85)80029-8

✦ Synopsis

The development of assistance and instructional systems is an alternative design strategy for human-computer interfaces. An integral component in the construction of these interfaces concerns the representation used to capture the user, task or machine. Representational issues and strategies in assistance and instructional systems are discussed in the paper. These representational issues concern the use of symbolic and quantitative models, the development of performance and cognitive models, the distinction between skilled and unskilled models, the use of static and dynamic models, and the possible development of multiple representation strategic3. From these issues, a profile methodology was distilled that could be implemented as an assistant or tutor in a file-search environment based upon novice and expert differences. This profile method is a quantitative, performance-based approach that samples expert behaviour to define a skilled model of file search. The development and formative evaluation of the methodology used 16 novice and 16 expert subjects. File-search performance and strategies were analysed in a complex information-retrieval task using the profile methodology. The method proved to be sensitive to novice and expert differences in file search and also seemed to be capable of being implemented as a diagnostic device for assistance and instruction. Limitations and extensions of the profile methodology are discussed in terms of command sequence information, explanation and reasoning, strategic and planning resources, and possible learning and adaptive models.