A method of recognizing the root of an improved dependency tree for the Chinese patent literature

Abstract

Compared with ordinary text, patent text in Chinese often has more complex sentence structure and more ambiguity of multiple verbs, which brings more difficulties in patent machine translation. To deal with these problems, this paper presents an improved dependency tree and a method to recognize the root of this tree for Chinese-English patent machine translation. Based on the theory of Hierarchical Network of Concepts (the HNC theory), some semantic features are used in the recognition. Experiments show that the precision of the recognition is close to 85%.