Understanding of Natural Language Text for Diagram Drawing

A. Mukherjee and U. Garain (India)


Artificial intelligence, knowledge representation, natural language processing, automated diagram drawing


There are many scientific and engineering software tools available that aids in drawing complex diagrams (in computer) comprising basic geometric entities. These tools can be used to create geometric figures following step-by-step commands given by the user. However, such tools are not intelligent enough to read and understand textual description of a diagram and directly generate it in the computer. The software application reported in this paper attempts to simulate human approach for drawing a geometric figure implied by a school-level geometric problem stated in natural language English. This involves machine understanding of the problem-text and then automatically creating the graphic representation by triggering relevant computer graphics functions. The system uses a language comprehension module for syntactic, semantic and logical analysis of a problem-text and also a custom knowledge base GeometryNet [1], comprising classified knowledge about objects and concepts related to school-level geometry, for geometric interpretation of the problem. Finally, the aim of this paper is to indicate an implementable framework for automatic text to diagram conversion using ideas from several fields like artificial intelligence, knowledge representation, natural language processing, automated reasoning etc.

Important Links:

Go Back

Rotating Call For Paper Image