The use of pivot languages and word-alignment techniques over bilingual corpora has proved an effective approach for extracting paraphrases of words and short phrases. However, inherent ambiguities in the pivot language(s) can lead to inadequate paraphrases. We propose a novel approach that is able to extract paraphrases by pivoting through multiple languages while discriminating word senses in the input language, i.e., the language to be paraphrased. Text in the input language is annotated with "senses" in the form of foreign phrases obtained from bilingual parallel data and automatic word-alignment. This approach shows 62% relative improvement over previous work in generating paraphrases that are judged both more accurate and more fluent.
展开▼