| Download | - View final version: Machine translation reference-less evaluation using YiSi-2 with bilingual mappings of massive multilingual language model (PDF, 689 KiB)
|
|---|
| Author | Search for: Lo, Chi-Kiu1; Search for: Larkin, Samuel1 |
|---|
| Affiliation | - National Research Council of Canada. Digital Technologies
|
|---|
| Format | Text, Article |
|---|
| Conference | 5th Conference on Machine Translation (WMT 2020), November 19-20, 2020, (Held Online) |
|---|
| Abstract | We present a study on using YiSi-2 with massive multilingual pretrained language models for machine translation (MT) reference-less evaluation. Aiming at finding better semantic representation for semantic MT evaluation, we first test YiSi-2 with contextual embeddings extracted from different layers of two different pretrained models, multilingual BERT and XLM-RoBERTa. We also experiment with learning bilingual mappings that transform the vector subspace of the source language to be closer to that of the target language in the pretrained model to obtain more accurate cross-lingual semantic similarity representations. Our results show that YiSi-2's correlation with human direct assessment on translation quality is greatly improved by replacing multilingual BERT with XLM-RoBERTa and projecting the source embeddings into the target embedding space using a cross-lingual linear projection (CLP) matrix learnt from a small development set. |
|---|
| Publication date | 2020-11-20 |
|---|
| Publisher | Association for Computational Linguistics (ACL) |
|---|
| Licence | |
|---|
| In | |
|---|
| Language | English |
|---|
| Peer reviewed | Yes |
|---|
| Identifier | 2020.wmt-1.100 |
|---|
| Export citation | Export as RIS |
|---|
| Report a correction | Report a correction (opens in a new tab) |
|---|
| Record identifier | 635c72b8-0570-44bf-862f-cb8265ea2a0a |
|---|
| Record created | 2022-07-14 |
|---|
| Record modified | 2022-07-15 |
|---|
