Abstract: As the development of machine proof technology,how to make the existing geometry instructional softwares to understand the latent semantics of the geometry proposition,automatically generate the figure and give out the reasoning process,is becoming a new challenge．Although the existing methods can transform the geometry proposition into figure or other machine-readable forms to a certain extent,they do not consider the dependent relationship between simple clauses,so they can not avoid the redundancies and inconsistence between construction commands of adjacent simple clauses．To resolve this problem,this paper proposed a merging algorithm,and further integrated the merging algorithm into exist natural language interface for automatic geometry construction to verify its effectiveness．The experiment results suggest that the merging algorithm can enhance the accuracy of automatic geometry construction from 84.17% to 91.67%,and can improve the accuracy of geometry proposition understanding.

Key words: Geometry proposition,Merging algorithm,Geometry construction,Natural language