BLEU Kishore Papineni, Salim Roukos, Todd Ward and Wei-Jing Zhu. (2002) BLEU: a method for Automatic Evaluation of Machine Translation. ACL. MT ( ) MT

Size: px
Start display at page:

Download "BLEU Kishore Papineni, Salim Roukos, Todd Ward and Wei-Jing Zhu. (2002) BLEU: a method for Automatic Evaluation of Machine Translation. ACL. MT ( ) MT"

Transcription

1 4. 1

2 BLEU Kishore Papineni, Salim Roukos, Todd Ward and Wei-Jing Zhu. (2002) BLEU: a method for Automatic Evaluation of Machine Translation. ACL. MT ( ) MT ( ) BLEU 2

3 BLEU BLEU MT NTCIR-7 3

4 ngram 1 MT ngram MT 1. 1 It is a guide to action 2 which 3 ensures that the military 4always obyes the 5 commands 6 of the party. 2. It is to ensure the troops forever hearing the activity guidebook that party direct 1. 1 It is a guide to action that 3 ensures that the military will forever heed Party 5 commands. 2. It is the guiding principle 2 which guarantees the military forces 4 always being under the command 6 of the party. 3. It is the practical guide for the army 4 always to heed the directions 6 of the party. 4

5 1. ngram MT ngram 2. 5

6 ngram ngram = ngram MT ngram MT: the the the the the the the Ref1: The cat is on the mat. Ref2: There is a cat on the mat. MT the the Ref1 Ref2 1gram = 7 7 6

7 ngram ngram ngram P n = MT ngram MT: the the the the the the the Ref1: The cat is on the mat. Ref2: There is a cat on the mat. P 1 = 2 7 P 2 = 0 7

8 MT1: P 1 = = 0.94, P 2 = = 0.59 MT2: P 1 = 8 14 = 0.57, P 2 = 1 13 = 0.08 MT 1. 1 It is a guide to action 2 which 3 ensures that the military 4always obyes the 5 commands 6 of the party. 2. It is to ensure the troops forever hearing the activity guidebook that party direct 1. 1 It is a guide to action that 3 ensures that the military will forever heed Party 5 commands. 2. It is the guiding principle 2 which guarantees the military forces 4 always being under the command 6 of the party. 3. It is the practical guide for the army 4 always to heed the directions 6 of the party. 8

9 ngram P n = MT MT ngram ngram MT MT ngram ngram MT ngram MT ngram MT 9

10 ngram N 1 n=1 N log P n P n = ngram N = ngram ( 4 ) ngram 10

11 ngram P n MT ngram ngram MT P n MT ngram P 1 = 2/2, P 2 = 1/1 MT : of the 1: It is a guide to action that ensures that the military will forever heed Party commands. 2: It is the guiding principle which guarantees the military forces always being under the command of the party. 3: It is the practical guide for the army always to heed the directions of the party. 11

12 MT MT ( ) c = MT MT r = MT BP (brevity penalty) BP = 1 if c r exp(1 r/c) if c < r MT BP = 1 ( ) MT < BP < 1 12

13 BLEU BLEU = BP exp( N 1 n=1 N log P n) MT ngram BLEU BLEU... MT 13

14 BLEU 10 ( ) 2 10 ( ) (S1,S2,S3,H1,H2) H1 H2 H1 H2 1( ) 5( )

15 (paired t-test) u i s r(u, i, s) s r(u, i, s ) s s d(u, i, s, s ) = r(u, i, s) r(u, i, s ) m(s, s ) = u,i d(u, i, s, s ) v(s, s ) = 1 n u,i (d(u, i, s, s ) m(s, s )) 2 n = u,i 1. s s m(s, s ) = 0 0 t = m(s, s ) v(s,s ) n 1 t 15

16 BLEU BLEU 2 S S S H H S1 0 S2 m(s1, S2) S3 = S2+m(S3, S2) 16

17 BLEU Normalized score Bilingual Monolingual BLEU S1 S2 S3 H1 H2 0-1 BLEU 2 BLEU S3 H1 H1 H2 S2 S3 17

18 BLEU Monolingual Judgment monolingual.txt f(x) BLEU BLEU 18

19 BLEU Bilingual Judgment bilingual.txt f(x) BLEU BLEU 2 19

20 BLEU MT BLEU BLEU ngram 20

Rapp BLEU[10] [9] BLEU OrthoBLEU Rapp OrthoBLEU [9] OrthoBLEU OrthoBLEU ) ) ) 1) NTT Natural Language Research

Rapp BLEU[10] [9] BLEU OrthoBLEU Rapp OrthoBLEU [9] OrthoBLEU OrthoBLEU ) ) ) 1) NTT Natural Language Research RJ-008 Is Back Translation Really Unuseful? Validation of Back Translation from the Perspective of a Checking Method for Users Mai Miyabe Takashi Yoshino 1. [1, 2] [3] [4] 1 2 2 [3,5,6,7] [8, 9] 1: 2 3

More information

20169 3 4 5003 n=3,000 61.8% 38.2% n=3,000 20 7.3% 30 21.3% 40 34.8% 50 36.6% n=3,000 3.0% 2.0% 1.5% 12.1% 14.0% 41.4% 25.9% n=3,000 37.7% % 24.8% 28.8% 1.9% 3.1% 0.2% n=3,000 500 64.0% 500 1,000 31.3%

More information

-----------------------------------------------------------------------------------------1 --------------------------------------------------------------------------------------1 -------------------------------------------------------------------------------------1

More information

FX ) 2

FX ) 2 (FX) 1 1 2009 12 12 13 2009 1 FX ) 2 1 (FX) 2 1 2 1 2 3 2010 8 FX 1998 1 FX FX 4 1 1 (FX) () () 1998 4 1 100 120 1 100 120 120 100 20 FX 100 100 100 1 100 100 100 1 100 1 100 100 1 100 101 101 100 100

More information

() () () 200,000 160,000 120,000 80,000 40,000 3.3 144,688 43,867 3.1 162,624 52,254 170,934 171,246 172,183 3 2.8 2.6 57,805 61,108 65,035 3.5 3 2.5 2 1.5 1 0.5 0 0 2 7 12 17 22 10.1 12.7 17 22.3 73.4

More information

JA2008

JA2008 A1 1 10 vs 3 2 1 3 2 0 3 2 10 2 0 0 2 1 0 3 A2 3 11 vs 0 4 4 0 0 0 0 0 3 6 0 1 4 x 11 A3 5 4 vs 5 6 5 1 0 0 3 0 4 6 0 0 1 0 4 5 A4 7 11 vs 2 8 8 2 0 0 0 0 2 7 2 7 0 2 x 11 A5 9 5 vs 3 10 9 4 0 1 0 0 5

More information

B B 10 7 581 10 8 582 10 9 583 B B 10 11 585 10 12 586 B 10 10 584 B

B B 10 7 581 10 8 582 10 9 583 B B 10 11 585 10 12 586 B 10 10 584 B 10 1 575 10 12 586 B B 10 1 575 10 2 576 B B 10 4 578 10 5 579 10 3 577 B 10 6 580 B B B 10 7 581 10 8 582 10 9 583 B B 10 11 585 10 12 586 B 10 10 584 B 11 1 587 11 12 598 B B 11 1 587 11 2 588 11 3 589

More information

(資料2)第7回資料その1(ヒアリング概要)

(資料2)第7回資料その1(ヒアリング概要) 2 3 4 5 6 7 8 9 10 11 12 13 1 1 1 1 5 1 6 533 4 505 722 13 3325 475 1 2 3 13 10 31 1 1 1 (1) 1 (2) 2 (3) 3 (4) 4 5 5 6 7 8 8 8 9 11 11 12 13 14 15 16 19 (1) (2) (3) (1) (5 ) 1 (10 ) ( ) (2) 2 4 (3) 3 3,100

More information

-------------------------- ----------------------------------------------------- -------------------------------------------------------------- ----------------------------------------------------- --------------------------------------------------------------

More information

<4D F736F F D DEC8BC A95BD90AC E A982BA81698AB A B B4790DF90AB8EBE8AB FC89408A4F816A82CC93AE8CFC82C98AD682B782E9838C837C815B D

<4D F736F F D DEC8BC A95BD90AC E A982BA81698AB A B B4790DF90AB8EBE8AB FC89408A4F816A82CC93AE8CFC82C98AD682B782E9838C837C815B D 27 29 2 IT 1,234 1,447 2,130 1,200 3,043 4 3 75 75 70-74 -10 J00 J101 J110 J111 J118 J300 J302-304 J301 26,475,118 155,290,311 1,234 14,472,130 75,784,748 12,003,043 79,505,563 1 1.00% 0.62% 1.31% 9 12

More information

, , ,210 9, ,

, , ,210 9, , 2006 5 642 7 2,671 35 732 1,727 602 489 386 74 373 533 305 1,210 9,786 2004 1,024 43.7 16.4 2004 978.6 40.2 2003 1 2006 5 1997 1998 1999 774 3,492 11 2,603 35 843 5,118 1,686 476 358 2000 738 3,534 11

More information

2 3 4 5 6 7 8 9 10 11 Licca is always with you 12 13 14 16 17 18 19 20 21 22 23 24 25 26 27 28 30 30 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31

More information

Present Situation and Problems on Aseismic Design of Pile Foundation By H. Hokugo, F. Ohsugi, A. Omika, S. Nomura, Y. Fukuda Concrete Journal, Vol. 29

Present Situation and Problems on Aseismic Design of Pile Foundation By H. Hokugo, F. Ohsugi, A. Omika, S. Nomura, Y. Fukuda Concrete Journal, Vol. 29 Present Situation and Problems on Aseismic Design of Pile Foundation By H. Hokugo, F. Ohsugi, A. Omika, S. Nomura, Y. Fukuda Concrete Journal, Vol. 29, No. 8, pp. 4-12, Aug. 1986 Synopsis The pile foundation

More information

研究報告 2 自動評価法を用いた機械翻訳の定量的評価 越前谷博 ( 北海学園大学 ) 磯崎秀樹 ( 岡山県立大学 ) 目次 1. 自動評価法とは 2. 自動評価法における動向 Workshop on Statistical Machine Translationに参加して 3. 自動評価法 :APA

研究報告 2 自動評価法を用いた機械翻訳の定量的評価 越前谷博 ( 北海学園大学 ) 磯崎秀樹 ( 岡山県立大学 ) 目次 1. 自動評価法とは 2. 自動評価法における動向 Workshop on Statistical Machine Translationに参加して 3. 自動評価法 :APA 研究会報告 2 自動評価法を用いた機械翻訳の定量的評価 研究報告 2 自動評価法を用いた機械翻訳の定量的評価 越前谷博 ( 北海学園大学 ) 磯崎秀樹 ( 岡山県立大学 ) 目次 1. 自動評価法とは 2. 自動評価法における動向 Workshop on Statistical Machine Translationに参加して 3. 自動評価法 :APAC 4. 自動評価法 :RIBES 5. まとめ

More information

1 1 tf-idf tf-idf i

1 1 tf-idf tf-idf i 14 A Method of Article Retrieval Utilizing Characteristics in Newspaper Articles 1055104 2003 1 31 1 1 tf-idf tf-idf i Abstract A Method of Article Retrieval Utilizing Characteristics in Newspaper Articles

More information

33 1 5 6 15 7 3 9 1 2 3 1 9 20 1 3 2002 56 1 1791 4 1946 1948 1949 1972 1899 1907 1919 1928 1945 3 2 150 3 48 75 22 10 6 51 40 25 12 14 1 3 7 17 7 16 96 62 26 28 9a 1 53 11 8 173 185 2 54 1990 162 1983

More information

IPSJ SIG Technical Report Vol.2016-CE-137 No /12/ e β /α α β β / α A judgment method of difficulty of task for a learner using simple

IPSJ SIG Technical Report Vol.2016-CE-137 No /12/ e β /α α β β / α A judgment method of difficulty of task for a learner using simple 1 2 3 4 5 e β /α α β β / α A judgment method of difficulty of task for a learner using simple electroencephalograph Katsuyuki Umezawa 1 Takashi Ishida 2 Tomohiko Saito 3 Makoto Nakazawa 4 Shigeichi Hirasawa

More information

http://www.ike-dyn.ritsumei.ac.jp/ hyoo/wave.html 1 1, 5 3 1.1 1..................................... 3 1.2 5.1................................... 4 1.3.......................... 5 1.4 5.2, 5.3....................

More information

Test IV, March 22, 2016 6. Suppose that 2 n a n converges. Prove or disprove that a n converges. Proof. Method I: Let a n x n be a power series, which converges at x = 2 by the assumption. Applying Theorem

More information

  医 事 法        山口大法医 藤宮龍也

  医 事 法             山口大法医 藤宮龍也 2 ( If anything can go wrong, it will. Things go wrong in batches. MiniMax (mini) (max) ( ) - 134 135 105 149 1) 2) 3) cf. cf. - 160-214 cf. - 211 209,210-204 vs 35 37-19 217 218-202 203 cf. - ( ) 104

More information

Z B- B- PHP - - [ ] PHP New York Times, December,,. The origins of the Japan-U.S. War and Adm. Isoroku Yamamoto Katsuhiko MATSUKAWA Abstract There are huge amount of studies concerning the origins

More information

90 120.0 80 70 72.8 75.1 76.7 78.6 80.1 80.1 79.6 78.5 76.8 74.8 72.4 69.5 95.6% 66.4 100.0 60 80.0 50 40 60.0 30 48.3% 38.0% 40.0 20 10 10.4% 20.0 0 S60 H2 H7 H12 H17 H22 H27 H32 H37 H42 H47 H52 H57 0.0

More information

2017 (413812)

2017 (413812) 2017 (413812) Deep Learning ( NN) 2012 Google ASIC(Application Specific Integrated Circuit: IC) 10 ASIC Deep Learning TPU(Tensor Processing Unit) NN 12 20 30 Abstract Multi-layered neural network(nn) has

More information

1: A/B/C/D Fig. 1 Modeling Based on Difference in Agitation Method artisoc[7] A D 2017 Information Processing

1: A/B/C/D Fig. 1 Modeling Based on Difference in Agitation Method artisoc[7] A D 2017 Information Processing 1,a) 2,b) 3 Modeling of Agitation Method in Automatic Mahjong Table using Multi-Agent Simulation Hiroyasu Ide 1,a) Takashi Okuda 2,b) Abstract: Automatic mahjong table refers to mahjong table which automatically

More information

ÉXÅ[ÉpÅ[ÉRÉuÉâï\éÜ

ÉXÅ[ÉpÅ[ÉRÉuÉâï\éÜ class SCLE HELICOPTER SUPERCOR W-1W ODY KIT INSTRUCTION MNUL SUPERCOR Please read this manual in its entirety before attempting to assemble the helicopter. This manual explains the parts exclusive to SUPERCOR.

More information

1 1,200m ,482,620m 1 323,825m 1 205,988m 114,988m 60,000m 31,000m ,837m ,158,795m ,340m 1 48,006m 1 33,016m 1,125m

1 1,200m ,482,620m 1 323,825m 1 205,988m 114,988m 60,000m 31,000m ,837m ,158,795m ,340m 1 48,006m 1 33,016m 1,125m IV 39 7 10 167 1 3 9 2 10 79 1 1,200m 23 23 6 1 429 2,482,620m 1 323,825m 1 205,988m 114,988m 60,000m 31,000m 27 117,837m 3 2 426 2,158,795m 44 277,340m 1 48,006m 1 33,016m 1,125m 33,016m 4 14,990m 2 44

More information

00 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.... 0........ 0 0 0 0 0 0 0 0 0 0..0..........0 0 0 0 0 0 0 0 0 0 0.... 0........ 0 0 0 0 0 0 0 0 0 0... 0...... 0... 0 0 0 0 0 0..0 0... 0 0 0 0 0.0.....0.

More information

Abstract :

Abstract : 17 18 3 : 3604U079- Abstract : 1 3 1.1....................................... 4 1................................... 4 1.3.................................. 4 5.1..................................... 6.................................

More information

3 ( 9 ) ( 13 ) ( ) 4 ( ) (3379 ) ( ) 2 ( ) 5 33 ( 3 ) ( ) 6 10 () 7 ( 4 ) ( ) ( ) 8 3() 2 ( ) 9 81

3 ( 9 ) ( 13 ) ( ) 4 ( ) (3379 ) ( ) 2 ( ) 5 33 ( 3 ) ( ) 6 10 () 7 ( 4 ) ( ) ( ) 8 3() 2 ( ) 9 81 1 ( 1 8 ) 2 ( 9 23 ) 3 ( 24 32 ) 4 ( 33 35 ) 1 9 3 28 3 () 1 (25201 ) 421 5 ()45 (25338 )(2540 )(1230 ) (89 ) () 2 () 3 ( ) 2 ( 1 ) 3 ( 2 ) 4 3 ( 9 ) ( 13 ) ( ) 4 ( 43100 ) (3379 ) ( ) 2 ( ) 5 33 ( 3 )

More information

Fig. 1 Schematic construction of a PWS vehicle Fig. 2 Main power circuit of an inverter system for two motors drive

Fig. 1 Schematic construction of a PWS vehicle Fig. 2 Main power circuit of an inverter system for two motors drive An Application of Multiple Induction Motor Control with a Single Inverter to an Unmanned Vehicle Propulsion Akira KUMAMOTO* and Yoshihisa HIRANE* This paper is concerned with a new scheme of independent

More information

[2] 1. 2. 2 2. 1, [3] 2. 2 [4] 2. 3 BABOK BABOK(Business Analysis Body of Knowledge) BABOK IIBA(International Institute of Business Analysis) BABOK 7

[2] 1. 2. 2 2. 1, [3] 2. 2 [4] 2. 3 BABOK BABOK(Business Analysis Body of Knowledge) BABOK IIBA(International Institute of Business Analysis) BABOK 7 32 (2015 ) [2] Projects of the short term increase at present. In order to let projects complete without rework and delays, it is important that request for proposals (RFP) are written by reflecting precisely

More information

ASP英語科目群ALE Active Learning in English No 7. What activity do you think is needed in ALE for students to improve student s English ability? active listening a set of important words before every lecture

More information

3 4 3 2 4 1 4 2 4 2 1 3 1 1 4 1 1 16,000 14,000 12,000 W) S) RC) CB 10,000 8,000 6,000 4,000 2,000 0 12,000 11,500 11,000 10,500 10,000 9,500 9,000 550 540 530 520 510 500 490 480 470 460 450 2008 2009

More information

dy + P (x)y = Q(x) (1) dx dy dx = P (x)y + Q(x) P (x), Q(x) dy y dx Q(x) 0 homogeneous dy dx = P (x)y 1 y dy = P (x) dx log y = P (x) dx + C y = C exp

dy + P (x)y = Q(x) (1) dx dy dx = P (x)y + Q(x) P (x), Q(x) dy y dx Q(x) 0 homogeneous dy dx = P (x)y 1 y dy = P (x) dx log y = P (x) dx + C y = C exp + P (x)y = Q(x) (1) = P (x)y + Q(x) P (x), Q(x) y Q(x) 0 homogeneous = P (x)y 1 y = P (x) log y = P (x) + C y = C exp{ P (x) } = C e R P (x) 5.1 + P (x)y = 0 (2) y = C exp{ P (x) } = Ce R P (x) (3) αy

More information

43 38 2000 6 1980 2004 8 24 2 8 26 9 2

43 38 2000 6 1980 2004 8 24 2 8 26 9 2 No.6, 42-53 (2005) One peculiar aspect of the military in North Korea MIYATA Atsushi Nihon University, Graduate school of Social and Cultural Studies North Korea is well-known for having one of the largest

More information

kut-paper-template.dvi

kut-paper-template.dvi 26 Discrimination of abnormal breath sound by using the features of breath sound 1150313 ,,,,,,,,,,,,, i Abstract Discrimination of abnormal breath sound by using the features of breath sound SATO Ryo

More information

lagged behind social progress. During the wartime Chonaikai did cooperate with military activities. But it was not Chonaikai alone that cooperated. Al

lagged behind social progress. During the wartime Chonaikai did cooperate with military activities. But it was not Chonaikai alone that cooperated. Al The Development of Chonaikai in Tokyo before The Last War Hachiro Nakamura The urban neighborhood association in Japan called Chonaikai has been more often than not criticized by many social scientists.

More information

IPSJ SIG Technical Report Pitman-Yor 1 1 Pitman-Yor n-gram A proposal of the melody generation method using hierarchical pitman-yor language model Aki

IPSJ SIG Technical Report Pitman-Yor 1 1 Pitman-Yor n-gram A proposal of the melody generation method using hierarchical pitman-yor language model Aki Pitman-Yor Pitman-Yor n-gram A proposal of the melody generation method using hierarchical pitman-yor language model Akira Shirai and Tadahiro Taniguchi Although a lot of melody generation method has been

More information

論文9.indd

論文9.indd Recent topics in surround sound audio Masaki Sawaguchi Summary This paper will describe the recent trends in surround sound system based on the results of various investigations on this topic. According

More information

Core Ethics Vol.

Core Ethics Vol. Core Ethics Vol. < > Core Ethics Vol. ( ) ( ) < > < > < > < > < > < > ( ) < > ( ) < > - ( ) < > < > < > < > < > < > < > < > ( ) Core Ethics Vol. ( ) ( ) ( ) < > ( ) < > ( ) < > ( ) < >

More information

- 2 -

- 2 - - 1 - - 2 - - 3 - - 4 - - 5 - ( ) - 6 - - 7 - - 8 - - 9 - - 10 - - 11 - - 12 - - 13 - - 14 - - 15 - - 16 - - 17 - What is your hobby? - 18 - - 19 - - 20 - - 21 - - 22 - - 23 - - 24 - - 25 - - 26 - - 27

More information