A Hybrid System for Chinese-English Patent Machine Translation

This paper presents a novel hybrid system, which combines rule-based machine translation (RBMT) with phrase-based statistical machine translation (SMT), to translate Chinese patent texts into English. The hybrid architecture is basically guided by the RBMT engine which processes source language parsing and transformation, generating proper syntactic trees for the target language. In the generation stage, the SMT subsystem then provides lexical translation according to the defined structures and generates final translation. According to our empirical evaluation, the hybrid approach outperforms each individual system across a varied set of automatic translation evaluation metrics, verifying the effectiveness of the proposed method.