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	NICT-5	INDICur-en	2018/08/24 15:01:22	2137	-	-	-	30.84	-	-	-	-	0.00	0.00	NMT	No	XX-XX transformer model
2	NICT-5	INDICur-en	2018/08/24 14:52:35	2119	-	-	-	27.88	-	-	-	-	0.00	0.00	NMT	No	XX-En transformer model
3	NICT-5	INDICur-en	2018/09/07 14:33:02	2243	-	-	-	26.73	-	-	-	-	0.00	0.00	NMT	No	Unified Source vocabulary by orthography mapping. XX-EN model.
4	IITP-MT	INDICur-en	2018/09/14 19:57:55	2351	-	-	-	26.56	-	-	-	-	0.00	0.00	NMT	No	Transformer multilingual XX-En
5	cvit	INDICur-en	2019/03/23 12:53:22	2665	-	-	-	25.57	-	-	-	-	0.00	0.00	NMT	Yes	massive-multi e270
6	cvit	INDICur-en	2019/03/14 22:12:49	2628	-	-	-	24.60	-	-	-	-	0.00	0.00	NMT	Yes	massive-multi + ft
7	cvit	INDICur-en	2019/03/22 05:47:10	2657	-	-	-	24.59	-	-	-	-	0.00	0.00	NMT	Yes	may to en (Transformer) - detokenized
8	cvit	INDICur-en	2019/03/22 05:32:54	2651	-	-	-	24.36	-	-	-	-	0.00	0.00	NMT	No	many to en (Transformer)
9	cvit	INDICur-en	2019/03/14 21:55:15	2621	-	-	-	21.03	-	-	-	-	0.00	0.00	NMT	Yes	massive-multi
10	NICT-5	INDICur-en	2018/08/24 14:52:20	2118	-	-	-	20.65	-	-	-	-	0.00	0.00	NMT	No	Bilingual transformer model
11	ORGANIZER	INDICur-en	2018/08/24 14:40:53	2103	-	-	-	19.60	-	-	-	-	0.00	0.00	NMT	No	multi2one multilingual NMT with Attention
12	ORGANIZER	INDICur-en	2018/08/29 14:35:46	2197	-	-	-	18.69	-	-	-	-	0.00	0.00	NMT	No	multi2multi multilingual NMT with Attention
13	Anuvaad	INDICur-en	2018/09/15 18:06:08	2411	-	-	-	18.31	-	-	-	-	0.00	0.00	SMT	No	SMT XX-En
14	Anuvaad	INDICur-en	2018/09/15 17:51:20	2401	-	-	-	18.03	-	-	-	-	0.00	0.00	SMT	No	SMT with KenLM
15	RGNLP	INDICur-en	2018/09/15 02:54:01	2379	-	-	-	15.36	-	-	-	-	0.00	0.00	SMT	No	SMT system with SRILM Language model
16	RGNLP	INDICur-en	2018/09/15 02:42:14	2372	-	-	-	14.88	-	-	-	-	0.00	0.00	SMT	No	SMT system with KENLM Language model
17	RGNLP	INDICur-en	2018/09/15 03:19:36	2386	-	-	-	10.65	-	-	-	-	0.00	0.00	NMT	No	NMT system with a 2-layer LSTM method
18	ORGANIZER	INDICur-en	2018/08/20 11:25:27	2012	-	-	-	9.29	-	-	-	-	0.00	0.00	NMT	No	NMT with Attention

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	cvit	INDICur-en	2019/03/22 05:47:10	2657	-	-	-	0.751459	-	-	-	-	0.000000	0.000000	NMT	Yes	may to en (Transformer) - detokenized
2	cvit	INDICur-en	2019/03/22 05:32:54	2651	-	-	-	0.746742	-	-	-	-	0.000000	0.000000	NMT	No	many to en (Transformer)
3	cvit	INDICur-en	2019/03/23 12:53:22	2665	-	-	-	0.744835	-	-	-	-	0.000000	0.000000	NMT	Yes	massive-multi e270
4	NICT-5	INDICur-en	2018/08/24 15:01:22	2137	-	-	-	0.742376	-	-	-	-	0.000000	0.000000	NMT	No	XX-XX transformer model
5	cvit	INDICur-en	2019/03/14 22:12:49	2628	-	-	-	0.741251	-	-	-	-	0.000000	0.000000	NMT	Yes	massive-multi + ft
6	IITP-MT	INDICur-en	2018/09/14 19:57:55	2351	-	-	-	0.733161	-	-	-	-	0.000000	0.000000	NMT	No	Transformer multilingual XX-En
7	cvit	INDICur-en	2019/03/14 21:55:15	2621	-	-	-	0.719253	-	-	-	-	0.000000	0.000000	NMT	Yes	massive-multi
8	ORGANIZER	INDICur-en	2018/08/24 14:40:53	2103	-	-	-	0.714075	-	-	-	-	0.000000	0.000000	NMT	No	multi2one multilingual NMT with Attention
9	NICT-5	INDICur-en	2018/08/24 14:52:35	2119	-	-	-	0.705470	-	-	-	-	0.000000	0.000000	NMT	No	XX-En transformer model
10	NICT-5	INDICur-en	2018/09/07 14:33:02	2243	-	-	-	0.696801	-	-	-	-	0.000000	0.000000	NMT	No	Unified Source vocabulary by orthography mapping. XX-EN model.
11	ORGANIZER	INDICur-en	2018/08/29 14:35:46	2197	-	-	-	0.695015	-	-	-	-	0.000000	0.000000	NMT	No	multi2multi multilingual NMT with Attention
12	Anuvaad	INDICur-en	2018/09/15 18:06:08	2411	-	-	-	0.635688	-	-	-	-	0.000000	0.000000	SMT	No	SMT XX-En
13	NICT-5	INDICur-en	2018/08/24 14:52:20	2118	-	-	-	0.631710	-	-	-	-	0.000000	0.000000	NMT	No	Bilingual transformer model
14	Anuvaad	INDICur-en	2018/09/15 17:51:20	2401	-	-	-	0.630810	-	-	-	-	0.000000	0.000000	SMT	No	SMT with KenLM
15	RGNLP	INDICur-en	2018/09/15 03:19:36	2386	-	-	-	0.619906	-	-	-	-	0.000000	0.000000	NMT	No	NMT system with a 2-layer LSTM method
16	RGNLP	INDICur-en	2018/09/15 02:54:01	2379	-	-	-	0.617035	-	-	-	-	0.000000	0.000000	SMT	No	SMT system with SRILM Language model
17	ORGANIZER	INDICur-en	2018/08/20 11:25:27	2012	-	-	-	0.611354	-	-	-	-	0.000000	0.000000	NMT	No	NMT with Attention
18	RGNLP	INDICur-en	2018/09/15 02:42:14	2372	-	-	-	0.605664	-	-	-	-	0.000000	0.000000	SMT	No	SMT system with KENLM Language model

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	NICT-5	INDICur-en	2018/08/24 15:01:22	2137	-	-	-	0.595450	-	-	-	-	0.000000	0.000000	NMT	No	XX-XX transformer model
2	IITP-MT	INDICur-en	2018/09/14 19:57:55	2351	-	-	-	0.583910	-	-	-	-	0.000000	0.000000	NMT	No	Transformer multilingual XX-En
3	cvit	INDICur-en	2019/03/23 12:53:22	2665	-	-	-	0.578870	-	-	-	-	0.000000	0.000000	NMT	Yes	massive-multi e270
4	cvit	INDICur-en	2019/03/22 05:32:54	2651	-	-	-	0.571530	-	-	-	-	0.000000	0.000000	NMT	No	many to en (Transformer)
5	cvit	INDICur-en	2019/03/22 05:47:10	2657	-	-	-	0.571530	-	-	-	-	0.000000	0.000000	NMT	Yes	may to en (Transformer) - detokenized
6	cvit	INDICur-en	2019/03/14 22:12:49	2628	-	-	-	0.566190	-	-	-	-	0.000000	0.000000	NMT	Yes	massive-multi + ft
7	NICT-5	INDICur-en	2018/09/07 14:33:02	2243	-	-	-	0.561620	-	-	-	-	0.000000	0.000000	NMT	No	Unified Source vocabulary by orthography mapping. XX-EN model.
8	NICT-5	INDICur-en	2018/08/24 14:52:35	2119	-	-	-	0.555430	-	-	-	-	0.000000	0.000000	NMT	No	XX-En transformer model
9	cvit	INDICur-en	2019/03/14 21:55:15	2621	-	-	-	0.544170	-	-	-	-	0.000000	0.000000	NMT	Yes	massive-multi
10	Anuvaad	INDICur-en	2018/09/15 17:51:20	2401	-	-	-	0.541890	-	-	-	-	0.000000	0.000000	SMT	No	SMT with KenLM
11	RGNLP	INDICur-en	2018/09/15 02:54:01	2379	-	-	-	0.541650	-	-	-	-	0.000000	0.000000	SMT	No	SMT system with SRILM Language model
12	RGNLP	INDICur-en	2018/09/15 02:42:14	2372	-	-	-	0.537230	-	-	-	-	0.000000	0.000000	SMT	No	SMT system with KENLM Language model
13	ORGANIZER	INDICur-en	2018/08/24 14:40:53	2103	-	-	-	0.532100	-	-	-	-	0.000000	0.000000	NMT	No	multi2one multilingual NMT with Attention
14	ORGANIZER	INDICur-en	2018/08/29 14:35:46	2197	-	-	-	0.523780	-	-	-	-	0.000000	0.000000	NMT	No	multi2multi multilingual NMT with Attention
15	Anuvaad	INDICur-en	2018/09/15 18:06:08	2411	-	-	-	0.519810	-	-	-	-	0.000000	0.000000	SMT	No	SMT XX-En
16	NICT-5	INDICur-en	2018/08/24 14:52:20	2118	-	-	-	0.481800	-	-	-	-	0.000000	0.000000	NMT	No	Bilingual transformer model
17	RGNLP	INDICur-en	2018/09/15 03:19:36	2386	-	-	-	0.468360	-	-	-	-	0.000000	0.000000	NMT	No	NMT system with a 2-layer LSTM method
18	ORGANIZER	INDICur-en	2018/08/20 11:25:27	2012	-	-	-	0.427350	-	-	-	-	0.000000	0.000000	NMT	No	NMT with Attention

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