A Robust Method for Understanding NL Interface Command in the ICI Intelligent Command Interpreter

To save computer users from the necessity of consulting computer systems manuals, and to make computer systems available to naive people, high-level user friendliness in computer systems becomes mandatory. ICI (Intelligent Command Interpreter) is an intelligent, natural language interface that serves as a computer systems consultant. This paper shows the mechanisms used by ICI to understand users' intentions and demands for commands. A command knowledge base and the results of a generalized LR parser based on unification-based grammars are used to understand users' demands. Intention links and an intention detection algorithm serve to understand users' intentions and anticipate their demands. ICI can act as needed to assist users in the accomplishment of their goals. ICI effectively increases user's productivity due to it keeps users from consulting written manuals. ICI also allows naive users to make efficient usage of computer systems, without the necessity of previous knowledge about systems usage.