Interested in Contributing?
- Check out available resources.
- Create an account and start submitting your own systems.
Scored Systems
| System | Submitter | System Notes | Run Notes | BLEU | BLEU-cased | TER |
|---|---|---|---|---|---|---|
| Google EN->CS (Details) | obo Charles University in Prague, UFAL |
as of Mar 1, 2010 | primary |
16.7 |
16.3 |
0.746 |
| uedin-wmt10-en-cz (Details) | Edinburgh University of Edinburgh |
constraint, no LDC data |
15.9 |
15.4 |
0.752 |
|
| CU Moses EN->CS WMT10 (Details) | obo Charles University in Prague, UFAL |
not constrained: uses large Czech LM and a little bit of more parallel data | primary |
15.7 |
14.9 |
0.745 |
| CU Moses EN->CS WMT10 (Details) | obo Charles University in Prague, UFAL |
not constrained: uses large Czech LM and a little bit of more parallel data | contrast (+Navajo in parallel, -smaller LM, 5gr) |
15.7 |
14.9 |
0.745 |
| CU Moses EN->CS WMT10 (Details) | obo Charles University in Prague, UFAL |
not constrained: uses large Czech LM and a little bit of more parallel data | contrast (best in SemPOS on newstest-2009) |
15.5 |
14.7 |
0.748 |
| uk-dan (Details) | zeman Charles University in Prague, ÚFAL |
Secondary run. Ondřej's maximal parallel corpus, hexagram LM. Problems with MERT, used weights from the primary run. |
13.4 |
12.6 |
0.749 |
|
| DCU_En--Cz_Primary system (Details) | DCU CNGL, Dublin City University |
System Combination of PB-SMT and Context-informed PB-SMT. | CNGL, School of Computing, Dublin City University, English to Czech PRIMARY SUBMISSION (constrained track) |
13.2 |
12.1 |
0.780 |
| CU TectoMT (Details) | popel UFAL, Charles University in Prague |
primary |
12.6 |
12.0 |
0.769 |
|
| uk-dan (Details) | zeman Charles University in Prague, ÚFAL |
Trained on 126144 sentence pairs of Czeng. 6gram LM trained on 13M Czech sentences. |
12.3 |
11.7 |
0.779 |
|
| UniPotsdam En-Cz (Details) | UniPotsdam Department of Linguistics, Potsdam University |
Constrained, no extra data, no extra LM |
12.3 |
12.0 |
0.799 |
|
| KU_TRBMT_L1 (Details) | ebicici Koc University |
L1 Regularized Transductive Regression Based Machine Translation reranking of Moses output |
11.7 |
10.5 |
0.821 |
|
| Microsoft Bing (Details) | obo Charles University in Prague, UFAL |
Run by a student of Ondrej, March 26, 2010. | primary, after deadline |
11.7 |
11.2 |
0.812 |
| SFU (Details) | baskarans Simon Fraser University |
In-house implementation of Hiero system | Constrained; LM generated using partial data |
11.3 |
10.9 |
0.814 |
| PC Translator (Details) | obo Charles University in Prague, UFAL |
a Czech commercial MT system | primary |
10.3 |
10.0 |
0.824 |
| Eurotran (Details) | obo Charles University in Prague, UFAL |
a Czech commercial MT system | primary |
9.9 |
9.6 |
0.819 |
| PC Translator (Details) | obo Charles University in Prague, UFAL |
a Czech commercial MT system | contrast (PC Translator 2007) |
9.4 |
9.1 |
failed |
| Google EN->CS (Details) | obo Charles University in Prague, UFAL |
as of Mar 1, 2010 | primary |
failed |
failed |
0.746 |
| CU Moses EN->CS WMT10 (Details) | obo Charles University in Prague, UFAL |
not constrained: uses large Czech LM and a little bit of more parallel data | contrast (larger mix 6-gr LM) |
failed |
failed |
0.745 |
| Eurotran (Details) | obo Charles University in Prague, UFAL |
a Czech commercial MT system | primary |
failed |
failed |
0.819 |
| CU Moses EN->CS WMT10 (Details) | obo Charles University in Prague, UFAL |
not constrained: uses large Czech LM and a little bit of more parallel data | primary (hopefully I'll still improve this tonight) |
failed |
failed |
0.748 |
| PC Translator (Details) | obo Charles University in Prague, UFAL |
a Czech commercial MT system | contrast (PC Translator 2007) |
failed |
failed |
failed |
| PC Translator (Details) | obo Charles University in Prague, UFAL |
a Czech commercial MT system | primary (PC Translator 2010) |
failed |
failed |
0.824 |