Quasinatural language in consultation systems

Following the development of question-answering systems, consultation systems have emerged as a useful integration between data-base management, knowledge representation, and natural-language interaction with the computer. The idea developed in this paper is to insert in a consultation system an input language which is formal but natural-oriented. That is a feasible input, which allows a natural communication with the computer without involving too many theoretical difficulties and practical disadvantages in terms of cost and execution time. The paper illustrates an implementation based on the classical facilities of the goal-oriented languages. Using microplanner as implementation language, we directly obtain the deep structure of the input quasinatural language expressed in microplanner, and this makes the deduction activity more reliable and direct. The problems related to knowledge representation and basic functions of the system are discussed; examples of its use are illustrated.