Minimally Supervised Method for Multilingual Paraphrase Extraction from Definition Sentences on the Web

We propose a minimally supervised method for multilingual paraphrase extraction from definition sentences on the Web. Hashimoto et al. (2011) extracted paraphrases from Japanese definition sentences on the Web, assuming that definition sentences defining the same concept tend to contain paraphrases. However, their method requires manually annotated data and is language dependent. We extend their framework and develop a minimally supervised method applicable to multiple languages. Our experiments show that our method is comparable to Hashimoto et al.’s for Japanese and outperforms previous unsupervised methods for English, Japanese, and Chinese, and that our method extracts 10,000 paraphrases with 92% precision for English, 82.5% precision for Japanese, and 82% precision for Chinese.