WAT

The Workshop on Asian Translation

Evaluation Results

BLEU

#	Team	Task	Date/Time	DataID	BLEU										Method	Other Resources	System Description
#	Team	Task	Date/Time	DataID	juman	kytea	mecab	moses- tokenizer	stanford- segmenter- ctb	stanford- segmenter- pku	indic- tokenizer	unuse	myseg	kmseg	Method	Other Resources	System Description
1	Bering Lab	JPCNko-ja	2021/04/25 02:58:43	5512	75.68	76.35	75.82	-	-	-	-	-	-	-	NMT	Yes	Transformer Ensemble with additional crawled parallel corpus
2	TMU	JPCNko-ja	2021/04/14 19:40:06	5151	73.67	74.53	73.92	-	-	-	-	-	-	-	NMT	No	Japanese BART, ensemble of 3 models
3	TMU	JPCNko-ja	2020/09/17 22:35:27	3829	73.13	74.03	73.45	-	-	-	-	-	-	-	NMT	No	Transformer, domain adaptation (BERT Japanese), shared EMB, shared BPE, ensemble of 4 models
4	TMU	JPCNko-ja	2020/09/17 22:38:16	3830	73.10	73.99	73.41	-	-	-	-	-	-	-	NMT	No	Transformer, domain adaptation (BERT Japanese), hanja loss, shared EMB, shared BPE, ensemble of 4 models
5	TMU	JPCNko-ja	2020/10/13 09:38:06	4137	73.07	73.95	73.38	-	-	-	-	-	-	-	NMT	No	Transformer, domain adaptation (BERT Korean), hanja loss, shared EMB, shared BPE, ensemble of 4 models
6	TMU	JPCNko-ja	2020/10/13 09:51:57	4141	73.06	73.97	73.39	-	-	-	-	-	-	-	NMT	No	Transformer, domain adaptation (BERT Korean), shared EMB, shared BPE, ensemble of 4 models
7	KNU_Hyundai	JPCNko-ja	2019/07/26 13:26:52	2998	73.04	73.95	73.32	-	-	-	-	-	-	-	NMT	No	Transformer + Relative Position + ensemble of 11 models
8	sarah	JPCNko-ja	2019/07/25 12:58:28	2924	72.55	73.44	72.83	-	-	-	-	-	-	-	NMT	No	Transformer, ensemble of 4 models
9	goku20	JPCNko-ja	2020/09/18 17:17:24	3918	72.20	73.07	72.48	-	-	-	-	-	-	-	NMT	No	Transformer, ensemble of 3 models
10	ryan	JPCNko-ja	2019/07/24 15:48:35	2890	71.78	72.67	72.08	-	-	-	-	-	-	-	NMT	No	Base Transformer
11	EHR	JPCNko-ja	2018/08/31 18:58:50	2214	71.70	72.68	72.01	-	-	-	-	-	0.00	0.00	NMT	No	SMT reranked NMT
12	sarah	JPCNko-ja	2019/07/22 12:44:02	2794	71.41	72.33	71.71	-	-	-	-	-	-	-	NMT	No	Transformer, single model
13	ORGANIZER	JPCNko-ja	2018/08/16 16:21:17	1997	70.13	70.97	70.39	-	-	-	-	-	0.00	0.00	NMT	No	NMT with Attention
14	goku20	JPCNko-ja	2020/09/18 17:28:33	3927	69.37	70.41	69.68	-	-	-	-	-	-	-	NMT	No	mBART pre-training, ensemble of 3 models
15	tpt_wat	JPCNko-ja	2021/04/27 01:54:36	5698	63.49	64.45	63.88	-	-	-	-	-	-	-	NMT	No	Base Transformer model with joint vocab, size 8k

Notice:

This table is sorted by the leftmost segmenters. You can change the segmenter used to sort by clicking each segmenter link.

RIBES

#	Team	Task	Date/Time	DataID	RIBES										Method	Other Resources	System Description
#	Team	Task	Date/Time	DataID	juman	kytea	mecab	moses- tokenizer	stanford- segmenter- ctb	stanford- segmenter- pku	indic- tokenizer	unuse	myseg	kmseg	Method	Other Resources	System Description
1	Bering Lab	JPCNko-ja	2021/04/25 02:58:43	5512	0.952025	0.951729	0.951775	-	-	-	-	-	-	-	NMT	Yes	Transformer Ensemble with additional crawled parallel corpus
2	TMU	JPCNko-ja	2021/04/14 19:40:06	5151	0.951822	0.951240	0.951721	-	-	-	-	-	-	-	NMT	No	Japanese BART, ensemble of 3 models
3	TMU	JPCNko-ja	2020/09/17 22:38:16	3830	0.950791	0.950247	0.950528	-	-	-	-	-	-	-	NMT	No	Transformer, domain adaptation (BERT Japanese), hanja loss, shared EMB, shared BPE, ensemble of 4 models
4	TMU	JPCNko-ja	2020/10/13 09:38:06	4137	0.950721	0.950179	0.950498	-	-	-	-	-	-	-	NMT	No	Transformer, domain adaptation (BERT Korean), hanja loss, shared EMB, shared BPE, ensemble of 4 models
5	TMU	JPCNko-ja	2020/09/17 22:35:27	3829	0.950506	0.949816	0.950269	-	-	-	-	-	-	-	NMT	No	Transformer, domain adaptation (BERT Japanese), shared EMB, shared BPE, ensemble of 4 models
6	TMU	JPCNko-ja	2020/10/13 09:51:57	4141	0.950401	0.949728	0.950233	-	-	-	-	-	-	-	NMT	No	Transformer, domain adaptation (BERT Korean), shared EMB, shared BPE, ensemble of 4 models
7	sarah	JPCNko-ja	2019/07/25 12:58:28	2924	0.949897	0.949399	0.949706	-	-	-	-	-	-	-	NMT	No	Transformer, ensemble of 4 models
8	goku20	JPCNko-ja	2020/09/18 17:17:24	3918	0.949838	0.949177	0.949627	-	-	-	-	-	-	-	NMT	No	Transformer, ensemble of 3 models
9	KNU_Hyundai	JPCNko-ja	2019/07/26 13:26:52	2998	0.949304	0.948709	0.949060	-	-	-	-	-	-	-	NMT	No	Transformer + Relative Position + ensemble of 11 models
10	EHR	JPCNko-ja	2018/08/31 18:58:50	2214	0.948529	0.947468	0.948416	-	-	-	-	-	0.000000	0.000000	NMT	No	SMT reranked NMT
11	ryan	JPCNko-ja	2019/07/24 15:48:35	2890	0.947986	0.947390	0.947783	-	-	-	-	-	-	-	NMT	No	Base Transformer
12	goku20	JPCNko-ja	2020/09/18 17:28:33	3927	0.947397	0.946878	0.947266	-	-	-	-	-	-	-	NMT	No	mBART pre-training, ensemble of 3 models
13	sarah	JPCNko-ja	2019/07/22 12:44:02	2794	0.947254	0.946995	0.947183	-	-	-	-	-	-	-	NMT	No	Transformer, single model
14	ORGANIZER	JPCNko-ja	2018/08/16 16:21:17	1997	0.941950	0.941336	0.941905	-	-	-	-	-	0.000000	0.000000	NMT	No	NMT with Attention
15	tpt_wat	JPCNko-ja	2021/04/27 01:54:36	5698	0.932805	0.931272	0.931897	-	-	-	-	-	-	-	NMT	No	Base Transformer model with joint vocab, size 8k

Notice:

This table is sorted by the leftmost segmenters. You can change the segmenter used to sort by clicking each segmenter link.

AMFM

#	Team	Task	Date/Time	DataID	AMFM										Method	Other Resources	System Description
#	Team	Task	Date/Time	DataID	unuse	unuse	unuse	unuse	unuse	unuse	unuse	unuse	unuse	unuse	Method	Other Resources	System Description
1	TMU	JPCNko-ja	2021/04/14 19:40:06	5151	0.935941	0.935941	0.935941	-	-	-	-	-	-	-	NMT	No	Japanese BART, ensemble of 3 models
2	Bering Lab	JPCNko-ja	2021/04/25 02:58:43	5512	0.930319	0.930319	0.930319	-	-	-	-	-	-	-	NMT	Yes	Transformer Ensemble with additional crawled parallel corpus
3	tpt_wat	JPCNko-ja	2021/04/27 01:54:36	5698	0.902731	0.902731	0.902731	-	-	-	-	-	-	-	NMT	No	Base Transformer model with joint vocab, size 8k
4	ORGANIZER	JPCNko-ja	2018/08/16 16:21:17	1997	0.000000	0.000000	0.000000	-	-	-	-	-	0.000000	0.000000	NMT	No	NMT with Attention
5	EHR	JPCNko-ja	2018/08/31 18:58:50	2214	0.000000	0.000000	0.000000	-	-	-	-	-	0.000000	0.000000	NMT	No	SMT reranked NMT
6	sarah	JPCNko-ja	2019/07/22 12:44:02	2794	0.000000	0.000000	0.000000	-	-	-	-	-	-	-	NMT	No	Transformer, single model
7	ryan	JPCNko-ja	2019/07/24 15:48:35	2890	0.000000	0.000000	0.000000	-	-	-	-	-	-	-	NMT	No	Base Transformer
8	sarah	JPCNko-ja	2019/07/25 12:58:28	2924	0.000000	0.000000	0.000000	-	-	-	-	-	-	-	NMT	No	Transformer, ensemble of 4 models
9	KNU_Hyundai	JPCNko-ja	2019/07/26 13:26:52	2998	0.000000	0.000000	0.000000	-	-	-	-	-	-	-	NMT	No	Transformer + Relative Position + ensemble of 11 models
10	TMU	JPCNko-ja	2020/09/17 22:35:27	3829	0.000000	0.000000	0.000000	-	-	-	-	-	-	-	NMT	No	Transformer, domain adaptation (BERT Japanese), shared EMB, shared BPE, ensemble of 4 models
11	TMU	JPCNko-ja	2020/09/17 22:38:16	3830	0.000000	0.000000	0.000000	-	-	-	-	-	-	-	NMT	No	Transformer, domain adaptation (BERT Japanese), hanja loss, shared EMB, shared BPE, ensemble of 4 models
12	goku20	JPCNko-ja	2020/09/18 17:17:24	3918	0.000000	0.000000	0.000000	-	-	-	-	-	-	-	NMT	No	Transformer, ensemble of 3 models
13	goku20	JPCNko-ja	2020/09/18 17:28:33	3927	0.000000	0.000000	0.000000	-	-	-	-	-	-	-	NMT	No	mBART pre-training, ensemble of 3 models
14	TMU	JPCNko-ja	2020/10/13 09:38:06	4137	0.000000	0.000000	0.000000	-	-	-	-	-	-	-	NMT	No	Transformer, domain adaptation (BERT Korean), hanja loss, shared EMB, shared BPE, ensemble of 4 models
15	TMU	JPCNko-ja	2020/10/13 09:51:57	4141	0.000000	0.000000	0.000000	-	-	-	-	-	-	-	NMT	No	Transformer, domain adaptation (BERT Korean), shared EMB, shared BPE, ensemble of 4 models

Notice:

This table is sorted by the leftmost segmenters. You can change the segmenter used to sort by clicking each segmenter link.
Adequacy-Fluency Metrics (AMFM) is a two-dimensional automatic evaluation metric for machine translation, designed to operate at the sentence level. It is based on adequacy and fluency, to decouple semantic and syntactic components of the translation process to provide a balanced view on translation quality.
AMFM is calculated without tokenizers.
The detail of AMFM is shown on the following paper: "Adequacy–Fluency Metrics: Evaluating MT in the Continuous Space Model Framework" [pdf]. Invited Talk in WAT2015 also helps understanding [slide].

HUMAN (WAT2022)

Notice:

HUMAN (WAT2022) is the result of the Pairwise Crowdsourcing Evaluation on WAT2022.
HUMAN (WAT2022) was evaluated by 5 different workers and the final decision is made by the voting of the judgements.

HUMAN (WAT2021)

Notice:

HUMAN (WAT2021) is the result of the Pairwise Crowdsourcing Evaluation on WAT2021.
HUMAN (WAT2021) was evaluated by 5 different workers and the final decision is made by the voting of the judgements.

HUMAN (WAT2020)

Notice:

HUMAN (WAT2020) is the result of the Pairwise Crowdsourcing Evaluation on WAT2020.
HUMAN (WAT2020) was evaluated by 5 different workers and the final decision is made by the voting of the judgements.

HUMAN (WAT2019)

Notice:

HUMAN (WAT2019) is the result of the Pairwise Crowdsourcing Evaluation on WAT2019.
HUMAN (WAT2019) was evaluated by 5 different workers and the final decision is made by the voting of the judgements.

HUMAN (WAT2018)

Notice:

HUMAN (WAT2018) is the result of the Pairwise Crowdsourcing Evaluation on WAT2018.
HUMAN (WAT2018) was evaluated by 5 different workers and the final decision is made by the voting of the judgements.

HUMAN (WAT2017)

Notice:

HUMAN (WAT2017) is the result of the Pairwise Crowdsourcing Evaluation on WAT2017.
HUMAN (WAT2017) was evaluated by 5 different workers and the final decision is made by the voting of the judgements.

HUMAN (WAT2016)

Notice:

HUMAN (WAT2016) is the result of the Pairwise Crowdsourcing Evaluation on WAT2016.
HUMAN (WAT2016) was evaluated by 5 different workers and the final decision is made by the voting of the judgements.

HUMAN (WAT2015)

Notice:

HUMAN (WAT2015) is the result of the Pairwise Crowdsourcing Evaluation on WAT2015.
HUMAN (WAT2015) was evaluated by 5 different workers and the final decision is made by the voting of the judgements.
The detail of the evaluation can be found in the pdf document (PDF file).

HUMAN (WAT2014)

Notice:

HUMAN (WAT2014) is the result of the Pairwise Crowdsourcing Evaluation on WAT2014.
HUMAN (WAT2014) was evaluated by 3 different workers and the final decision is made by the voting of the judgements.
The detail of the evaluation can be found in the pdf document (PDF file).

EVALUATION RESULTS USAGE POLICY

When you use the WAT evaluation results for any purpose such as:
- writing technical papers,
- making presentations about your system,
- advertising your MT system to the customers,
you can use the information about translation directions, scores (including both automatic and human evaluations) and ranks of your system among others. You can also use the scores of the other systems, but you MUST anonymize the other system's names. In addition, you can show the links (URLs) to the WAT evaluation result pages.

NICT (National Institute of Information and Communications Technology)
Kyoto University
Last Modified: 2018-08-02

WAT The Workshop on Asian Translation Evaluation Results

BLEU

RIBES

AMFM

HUMAN (WAT2022)

HUMAN (WAT2021)

HUMAN (WAT2020)

HUMAN (WAT2019)

HUMAN (WAT2018)

HUMAN (WAT2017)

HUMAN (WAT2016)

HUMAN (WAT2015)

HUMAN (WAT2014)

EVALUATION RESULTS USAGE POLICY

WAT

The Workshop on Asian Translation

Evaluation Results