ABSTRACT
Knowledge-based systems aim at interpreting the queries before providing an answer. This task is domain intensive and requires the use of a knowledge base tha t stores semantic information about the domain and also often about the English language itself. Examples of knowledge-based help systems are numerous, e.g., UC, the UNIX Consultant [Wilensky et al., 1984], INTERIX [Guez, 1987], N L H /E [Tichy et al., 1989], etc. Most knowledge-based sys tems do more than simply give pointers to tools or documentation. Some of them are context sensitive and generate answers adapted to the user’s exper tise. As a tradeoff, the encoding of the knowledge base is often tedious and expensive, especially for large toolkits. Moreover, the knowledge base is so domain dependent that a large part of the help system has to be rebuilt for each new domain of application.
