Using Restricted Natural Language for Geometric Construction

Abstract

Human-Computer Interaction is crucial for Instructional Software. It is significant to make Geometry Instructional Software understand geometric proposition described by natural language and construct geometric figure automatically. However, it is difficult for computer to accurately understand the flexible and variable geometric proposition because of the immaturity of Natural Language Processing Technology. This paper defines a restricted geometric proposition, presents a corresponding comprehension model based on ontology, and designs a matching algorithm for geometric relation pattern, finally transforms the restricted geometric proposition to geometric construction command sequence. The experiment results suggest that the average accuracy of understanding clauses of geometric proposition is almost to 88.9%, and the percentage of constructing correct geometric figure is to 82.5%.