Go Back
Data Set
The MetricsMATR 2008 evaluation data set is not to be publicly released. Portions will be reused for future NIST MT evaluations.
Primary Evaluation Set
| Origin | Source Language | Target Language | Genre(s) | Words (est.) | Systems |
| MT08 | Arabic | English | NW, WB | 15,000 | 10 |
| Chinese | English | NW, WB | 15,000 | 10 |
| GALE P2 | Arabic | English | NW, WB | 11,500 | 3 |
| Chinese | English | NW, WB | 10,000 | 3 |
| GALE P2.5 | Arabic | English | BN | 5,500 | 2 |
| Chinese | English | BC, BN | 10,000 | 3 |
| Transtac, Jul 07 | Arabic | English | Dialog | 6,500 | 5 |
| Farsi | English | Dialog | 4,500 | 5 |
| Transtac, Jan 07 | Arabic | English | Dialog | 5,000 | 5 |
Secondary Evaluation Set
| Origin | Source Language | Target Language | Genre(s) | Words (est.) | Systems |
| CESTA, run1 | Arabic | French | General | 28,000 | 2 |
| English | French | General | 21,500 | 5 |
| CESTA, run2 | Arabic | French | Health | 20,000 | 1 |
| English | French | Health | 22,500 | 5 |