Papers by Aleksandr Khakhmovich
Cross-lingual Named Entity List Search via Transliteration (2020.lrec-1)
Copied to clipboard
| Challenge: | a common way to adapt out-of-vocabulary words is a challenge in cross-lingual tasks . intrinsic evaluation, i.e comparison to a single gold standard, might not be appropriate in the task of transliteration due to its high variability. |
| Approach: | They propose to train Transformer-based multilingual transliteration models on 6 high- and 4 less-resourced languages and compare them with bilingual models. |
| Outcome: | The proposed model outperforms bilingual models on less-resourced languages. |