From planification to search in computer help systems

Existing literature on help systems reports on major difficulties novice users encounter when trying to learn new software functionalities through the use of an embedded help system. Far to be a simple and univocal task, information search in a help system is most often accomplished as a secondary task during a primary activity that is performed in a dynamic context under specific constraints (e.g. professional situations). We assume that one of the main sources of difficulty for users lies in the formulation of a definite search query from an ill-defined problem encountered in the primary activity. We carried out an experiment to evaluate the users’ performance and search strategies with different types of targets. The targets were defined as either simple or complex according to the familiarity of their denomination, and they could be reached in either 3 or 4 steps in the help system arborescence. In order to be as close as possible to an ecological text formatting task, the users were asked to reproduce a formatted text in which four elements were to be searched in the help system. The results showed that the familiarity of the target denomination affected dramatically the searching performance, but that the depth of its position in the help system arborescence had little effect on it. We conclude in deriving some guidelines for the design of help systems that would effectively support learning.