Automatic Generation of Context-Sensitive Textual Help

Abstract

Graphical user interfaces tend to get more and more complex, and consequently, the user needs help to work with these interfaces. Development of good help systems is very time-consuming, if only because the help system needs to change whenever the specifications do. Therefore, we have been concentrating on the automatic generation of help.

The context of the research reported in this paper is the User Interface Design Evvironment (UIDE), which uses high-level application and interface representations to generate interfaces and help. Based on descriptions of an interface, help can be generated to answer two questions: "Why is an object not usable (grayed out)?", and "How can it be made usable?". Pre- and post-conditions, part of the semantics and sequencing control representations in UIDE, are used to answer these questions.

The help system uses unsatisfied preconditions to explain why an object in the interface is disabled. A standard set of templates is used to translate often occurring predicates into text. A planner is used to derive how to enable the object in question. The help system looks at those predicates that must become true to enable the object, and searches for objects with corresponding post-conditions.