Learning Paraphrastic Sentence Embeddings from Back-Translated Bitext
This work addresses the need for scalable, high-quality training data for paraphrastic sentence embeddings, which is incremental by building on prior methods like Wieting et al. (2016b).
The paper tackled the problem of learning general-purpose paraphrastic sentence embeddings by generating sentential paraphrases via back-translation of bilingual sentence pairs, finding that the data quality is on par with manually-written English paraphrase pairs and can scale to many languages and domains.
We consider the problem of learning general-purpose, paraphrastic sentence embeddings in the setting of Wieting et al. (2016b). We use neural machine translation to generate sentential paraphrases via back-translation of bilingual sentence pairs. We evaluate the paraphrase pairs by their ability to serve as training data for learning paraphrastic sentence embeddings. We find that the data quality is stronger than prior work based on bitext and on par with manually-written English paraphrase pairs, with the advantage that our approach can scale up to generate large training sets for many languages and domains. We experiment with several language pairs and data sources, and develop a variety of data filtering techniques. In the process, we explore how neural machine translation output differs from human-written sentences, finding clear differences in length, the amount of repetition, and the use of rare words.