A computer-aided collection and construction system of terminology based on a statistically built knowledge base

Since a terminology is a system of concepts to which terms should be assigned, automatic processings of such concepts seem to be difficult to be realized. In this paper, we propose a computer-aided collection and construction system of terminology which creates automatically a knowledge base using the technical terms and their frequency of appearance information extracted from a bibliographic database and selects the terminologies based on the knowledge base. The experts in specialized fields can be offered a set of candidates for the technical terms with the system and determined needed terminologies through the following important procedures: 1) the relations among important concepts in a specified field can be extracted; 2) the concepts in the lower classes can be searched by specifying those in the upper classes; 3) the important terms can be distinguished from the basic terms and then the most important terms can be selected. An application of the present system to nuclear fusion research is described in detail. (author)