Interested in Contributing?

Scored Systems

System Submitter System Notes Constraint Run Notes BLEU BLEU (11b) BLEU-cased BLEU-cased (11b) TER z BLEU-cased-norm zz BEER 2.0 zzz CharactTER
abumatran-enfi-uncons-combo  (Details) atoral
Dublin City University
combination of unconstrained (unsegmented and rule-based compound segmented) and constrained (rule-based and unsupervised morph segmented) models no

16.0

16.0

15.5

15.5

0.777

abumatran-enfi-uncons  (Details) rrubino
Saarland University & DFKI
PB-SMT, OSM, 3 reordering models, additional parallel (FiEnWaC, OpenSubs) and monolingual (FiWaC) data no

15.3

15.3

14.9

14.9

0.803

UU-enfi-unconstrained  (Details) jorgtied
University of Helsinki
no phrase-based system with OPUS and crawled monolingual data

14.8

14.8

13.7

13.7

0.796

uedin-pbt-wmt15-en-fi  (Details) barry
University of Edinburgh
no Moses, Opus data, OSm

13.8

13.8

13.4

13.4

0.803

abumatran-enfi-combo  (Details) atoral
Dublin City University
combination of unsegmented and segmented models (rule-based and unsupervised) yes

13.0

13.0

12.7

12.7

0.804

CMU.fi-en  (Details) chris_dyer
Carnegie Mellon
yes

12.9

12.9

12.5

12.5

0.798

uedin-syntax-en-fi  (Details) Phil Williams
University of Edinburgh
Moses tree-to-string yes

12.8

12.8

12.3

12.3

0.809

abumatran-enfi-hfstcomp  (Details) rrubino
Saarland University & DFKI
HFST-COMP segmentation, PB-SMT, OSM, 3 reordering models yes

12.0

12.0

11.6

11.6

0.830

AaltoMorphsRescored  (Details) Stig-Arne Grönroos
Aalto University
Unsupervised morph segmentation, MERT, RNNLM rescoring. yes With RNNLM rescoring.

12.0

12.0

11.6

11.6

0.808

ParFDA5-DCU  (Details) bicici
Parallel Feature Decay Algorithms (ParFDA5) Moses RELEASE-3.0 phrase based SMT results. yes en-fi (after the deadline).

11.8

11.8

11.3

11.3

0.844

AaltoMorphs  (Details) Stig-Arne Grönroos
Aalto University
Unsupervised morph segmentation, MERT, without RNNLM rescoring. yes without RNNLM rescoring.

11.6

11.6

11.2

11.2

0.829

USAAR-gacha-ENFI  (Details) anything
uni saarland
yes gacha v2

10.8

10.8

10.5

10.5

0.850

ParFDA5-DCU  (Details) bicici
Parallel Feature Decay Algorithms (ParFDA5) Moses RELEASE-3.0 phrase based SMT results. yes en-fi, LM is train target.

10.7

10.7

10.3

10.3

0.856

UU-enfi  (Details) jorgtied
University of Helsinki
yes phrase-based baseline

10.7

10.7

9.8

9.8

0.842

USAAR-gacha-ENFI  (Details) anything
uni saarland
yes no gacha

9.9

9.9

9.6

9.6

0.873

GF Wide-coverage system  (Details) prakol
Chalmers University of Technology
The current system is a wide-coverage translation system developed using Grammatical Framework. The system is an Interlingua-based system. yes Baseline GF translator with rules for case-marker handling

4.8

4.8

4.6

4.6

1.135

GF Wide-coverage system  (Details) prakol
Chalmers University of Technology
The current system is a wide-coverage translation system developed using Grammatical Framework. The system is an Interlingua-based system. yes From top 1000 translations, scored using LM

4.8

4.8

4.6

4.6

1.135

GF Wide-coverage system  (Details) prakol
Chalmers University of Technology
The current system is a wide-coverage translation system developed using Grammatical Framework. The system is an Interlingua-based system. yes Baseline wide-coverage translator from English to Finnish

4.7

4.7

4.5

4.5

1.138

apertium-fin-eng-unconstrained-en-fi  (Details) tpirinen
Hamburger Zentrum für Sprachkorpora
This is RBMT with large monolingual and bilingual dictionaries but minimal amount of rules. This submission was made by "tuning" on dev sets, i.e. manually fixing system to be best on dev sets. no This is an out of the box rbmt with only minimal effort changes by adding dev and test set words missing from bilingual dictionary, but not from the monolingual ones, added. Lexical selection etc. was not tuned.

3.0

3.0

2.9

2.9

1.074