 |
ML Newsnew | past | best | rss | submit | about |
| Simple and Effective Paraphrastic Similarity from Parallel Translations (arxiv.org) | |
| 2 points by sel1 on Oct 1, 2019 | hide | past | pdf | discuss on HN | |
|
In plain words: The model learns sentence meanings straight from pairs of the same text in two languages, skipping the usual step of hand-building lists of reworded sentences. On tasks comparing text across languages, it beat more complicated systems while running orders of magnitude faster. AbstractWe present a model and methodology for learning paraphrastic sentence embeddings directly from bitext, removing the time-consuming intermediate step of creating paraphrase corpora. Further, we show that the resulting model can be applied to cross-lingual tasks where it both outperforms and is orders of magnitude faster than more complex state-of-the-art baselines. | |
| add comment on HN |