A knowledge-based framework in an intelligent assistant system for making documents

Abstract We have developed a prototype of an intelligent assistant system for making documents in a specific domain. For one example, the domain of Japanese business letters is chosen, because it contains such formal sentences that sophisticated writing skill is required. Then we have implemented the prototype system in Prolog. The system is designed to help novices make business letters efficiently by retrieving the relevant and associated paragraphs from a number of examples in a text database. In this paper, first the associative retrieval mechanism, which is realized by using a retrieval technique on the basis of keywords, is presented. And then, a realistic approach for a content extraction and documents summarizing method, which aims at content retrieval, is described.