Building A Parallel Corpus with Bilingual Discourse Alignment

摘要

This paper describes a discourse resource, namely a Chinese-English parallel corpus, based on the idea of bilingual discourse alignment. We introduce a bilingual collaborative annotation approach, which annotates English discourse units based on Chinese ones, and annotates Chinese discourse structure based on English ones subsequently. Such approach can ensure full discourse structure alignment between parallel texts, and reduce cost for annotating texts of two languages as well. Annotation Evaluation of the parallel corpus justifies the appropriateness of the discourse alignment framework to parallel texts.