Research on Element Sub-sentence in Chinese-English Patent Machine Translation

This paper presents an approach to translate element sub-sentences which widely exist in Chinese patent documents. Element sub-sentence is a kind of language chunk in a sentence in which one part of sub-sentence is the headword and others are attributives, or the modifier-head phrase of VP+NP or NP+VP structure. Element sub-sentence can be divided into three types in HNC theory: EK (predicate) sub-sentence, GBK1 (subject) sub-sentence and GBK2 (object) sub-sentence. In this paper, we give the method to detect the sub-sentence boundaries, analyze the characteristic of each type of element sub-sentences from the structure and semantics, and discover the Chinese-English translation rules of each type. By using the processing strategies most of the Chinese-English translation problem about the element sub-sentence can be perfectly solvable on an online patent MT system in SIPO.