$B $B$3$NJ8=q$K$D$$$F(B... $B>e$X(B: Statistical Machine Translation with $BLa$k(B: Acknowledgements

$BJ88%L\O?(B

1

Peter F. Brown, Stephen A. Della Pietra, Vincent J. Della Pietra, and Robert L. Mercer. ``The machinematics of machine translation: Parameter estimation'', Computational Linguistics, 19(2): pp. 263-311. (1993).

2

Philipp Koehn, Franz J. Och, and Daniel Marcu. ``Statistical phrase-based translation'', In Marti Hearst and Mari Ostendorf, editors, HLT-NAACL 2003: Main Proceedings, pages 127.133, Edmonton, Alberta, Canada, May 27 -June 1. Association for Computational Linguistics. (2003).

3

Pierre Isabelle, Cyril Goutte, and Michel Simard., ``Domain Adaptation of MT systems through automatic post-editing'', MT Summit XI, 102, 2007.

4

Yushi Xu and Stephanie Seneff, ``Two-Stage Translation: A Combined Linguistic and Statistical Machine Translation Framework'', Proceedings of the Eighth Conference of the Association for Machine Translation (AMTA) 2008.

5

Jason Katz-Brown and Michael Collins, ``Syntactic Reordering in Preprocessing for Japanese English Translation: MIT System Description for NTCIR-7 Patent Translation Task'', Proceedings of the 7th NTCIR Workshop Meeting, 2008.

5

Franz Josef Och, Hermann Ney. ``A Systematic Comparison of Various Statistical Alignment Models'', Computational Linguistics, volume 29, number 1, pp. 19-51 March 2003.

6

Andreas Stolcke, ``SRILM - An Extensible Language Modeling Toolkit'', in Proc. Intl. Conf. Spoken Language Processing, Denver, Colorado, September 2002

7

Philipp Koehn, Marcello Federico, Brooke Cowan, Richard Zens, Chris Dyer, Ond$B*T(Bej Bojar, Alexandra Constantin, Evan Herbst, ``Moses: Open Source Toolkit for Statistical Machine Translation'', Proceedings of the ACL 2007 Demo and Poster Sessions, pages 177$B#|(B180, June 2007.

8

NIST Open Machine Translation,
http://www.nist.gov/speech/tests/mt

9

Banerjee, S. and A. Lavie, ``METEOR: An Automatic Metric for MT Evaluation with Improved Correlation with Human Judgments'', Proceedings of Workshop on Intrinsic and Extrinsic Evaluation Measures for MT and/or Summarization at the 43th Annual Meeting of the Association of Computational Linguistics (ACL-2005), June 2005.

10

Matthew Snover, Bonnie Dorr, Richard Schwartz, Linnea Micciulla, and John Makhoul, ``A Study of Translation Edit Rate with Targeted Human Annotation,'' Proceedings of Association for Machine Translation in the Americas, 2006.

11

Papineni, K., Roukos, S., Ward, T., and Zhu, W. J., ``BLEU: a method for automatic evaluation of machine translation'' in ACL-2002: 40th Annual meeting of the Association for Computational Linguistics pp. 311-318, 2002.

12

Hideki Isozaki, Tsutomu Hirao, Kevin Duh, Katsuhito Sudoh, Hajime Tsukada, ``Automatic Evaluation of Translation Quality for Distant Language Pairs'', Conference on Empirical Methods on Natural Language Processing (EMNLP), Oct. 2010.

13

Hiroshi Echizen-ya, Kenji Araki, ``Automatic Evaluation of Machine Translation based on Recursive Acquisition of an Intuitive Common Parts Continuum'', Proceedings of the Eleventh Machine Translation Summit (MT SUMMIT XI), Page.151-158, Copenhagen, Denmark, 2007.

14

NAACL 2006 WORKSHOP ON STATISTICAL MACHINE TRANSLATION Shared Task: Exploiting Parallel Texts for Statistical Machine Translation Shared Task Baseline System, training-release-1.3.tgz, http://www.statmt.org/wmt06/shared-task/baseline.html

15

Jin'ichi Murakami, Masato Tokuhisa, Satoru Ikehara, ``Statistical Machine Translation adding Pattern-based Machine Translation in Chinese-English Translation'', International Workshop on Spoken Language Translation 2009, (IWSLT 2009), pp.107-112, October 2009.

16

L.Dugast, J.Senellart, and P.Koehn, ``Statistical postediting on SYSTRAN's rule-based translation system'', in Second Workshop on SMT, 2007, pages.179-182

17

M.Simard, N.Ueffing, P.Isabelle, and R.Kuhn, ``Rule-based translation with statistical phrase-based post-editing'', in Second Workshop on SMT, 2007, pages.203-206

18

Terumasa EHARA, ``Rule Based Machine Translation Combined with Statistical Post Editor for Japanese to English Patent Translation'', Proceedings of Machine Translation Summit XI, Workshop on Patent Translation, pp.13-18, Sept., 2007.

19

Terumasa EHARA, ``Machine Translation for Patent Documents Combining Rule-based Translation and Statistical Post-editing '',Proceedings of NTCIR-8 Workshop Meeting, pp.384-386, 2010.

20

Jin'ichi Murakami, Masato Tokuhisa, Satoru Ikehara, ``Statistical Machine Translation Adding Rule-based Machine Translation'', Proceedings of NTCIR-8 Workshop Meeting, pp.391-396, 2010.

$BI=(B 6: Outputs

J E
2	Input	$B?^#4$KHs<'@-AX$rN.$l$kEEN.$KBP$9$kEE5$Dq93$r=D<4$K$H$C$F%0%i%U$K<($7$F$$$k!#(B
	Proposed	The electric resistance to the current flowing in FIG . 4 via a non-magnetic layer is shown in the graph on the vertical axis .
	Baseline	FIG . 4 is a graph showing the electric current flowing through the resistance nonmagnetic layer vertical axis .
	RBMT	The electrical resistance to the current which flows into Fig. 4 through a nonmagnetic layer is shown in the graph for the vertical axis.
	REFERENCE	FIG . 4 is a graph showing an electric resistance to an electric current flowing in the nonmagnetic layer with the axis of ordinates representing the electric resistance .
3	Input	$B?^#6$O!"3,CJGH$NGH9bCM$r#0!'!^#1!'!^"e#2$N#5CM$G!"3n$D!"3,CJGH$NGH7AN)$,>e$j;~4V$rCY$/$7$?>l9g$N$b$N$G$"$k!#(B
	Proposed	FIG . 6 shows that obtained when five values of 0 : 1 : & plusmn ; 2 , and $BGH7AN)(B of a staircase waveform up the crest value of a staircase waveform , and the time delayed .
	Baseline	In step 5 0 , . + -.1 / 2 of the peak value of this waveform is delayed , and when the staircase waveform of FIG . 6 , is provided .
	RBMT	Fig. 6 shows the thing at the time of being five values of 0:$B!^(B1:$B!^(Broot2, and $BGH7AN)(B of a stairs wave going up the wave high price of a stairs wave, and making time late.
	REFERENCE	In FIG . 6 , the peak values of the staircase wave are set to five values , 0 , 1 , and 2 , and the wave is raised in delayed timings .
4	Input	$B$^$?!"%5!<%b%a!<%?%(%s%3!<%@#1#1$O!"F~NO?.9f#I#N#0$,!V#1#6!W!J#1#0?J?t!K$N>l9g$KNc30=hM}$r9T$$!">jM>%3!<%I#C#0$K!V#4!W$r=PNO$7!"%^%k%A%W%k%3!<%I#C#1#m#u#l#t#i#p#l#e$K!V#1#2!W$r=PNO$9$k!#(B
	Proposed	Further , the thermometer film encoder 1 1 conducts an exceptional process , when the input signal IN0 is " 1 6 " ( decimal number ) , it outputs " 4 " to the excessive code C0 and outputs " 1 2 " to multiple code C1 .
	Baseline	he remainder is " 0 " , " 1 6 " and " 4 " in decimal number ) , the output of the thermometer 1 2 $B#m#u#l#t#i#p#l#e(B exception code ( C1 code ) encoder 1 0 1 is subjected to the multiple outputs the input signal IN .
	RBMT	Moreover, the thermometer encoder 11 performs exception handling, when input signal IN0 is "16" (decimal number), it outputs "4" to the surplus code C0, and outputs "12" to multiple code C1multiple.
	REFERENCE	Also , when input signal IN0 is 1 6 ( decimal number ) , thermometer encoder 1 1 performs exceptional treatment , outputs 4 to remainder code C0 , and outputs 1 2 to multiple code C1 multiple .
14	Input	$B$=$7$F!"%3%M%/%?#1#8#a$r4pHD#1#2$K8GDj$9$k$H$-$K$O!"%3%M%/%?#1#8#a$Ne$K0LCV$5$;!J?^#3$b;2>H!K!"$=$N8e
	Proposed	When the connector 1 8 a is fixed to the substrate 1 2 , resin claws 3 0 of the connector 1 8 a is positioned on the pawl engaging hole 3 2 of the substrate 1 2 ( see FIG . 3 ) , and the resin claws 3 0 is inserted into the pawl engaging hole 3 2 .
	Baseline	The connector 2 3 is inserted into the connector 3 0 , whereupon the resin substrate 3 8 is placed on the resin substrate 1 2 2 of a pawl member 1 2 0 is fixed at the pawl engaging holes 1 3 a ( see also FIG . 3 ) of the pawl engaging holes 1 8 .
	RBMT	And when the connector 18a is fixed to the substrate 12, the resin nail 30 of the connector 18a is located on the nail engaging-of-clutch hole 32 of the substrate 12 (also see Fig. 3), and the resin nail 30 is inserted into the nail engaging-of-clutch hole 32 after that.
	REFERENCE	When the connector 1 8 a is fixed to the board 1 2 , the resin claw 3 0 of the connector 1 8 a is positioned above the claw engagement hole 3 2 of the board 1 2 ( also refer to FIG . 3 ) , and the resin claw 3 0 is thereafter inserted in the claw engagement hole 3 2 .
E J
1	Input	As shown in FIG. 4, the valve overlap amount decreases as the valve working angle of the intake valve 30 decreases.
	Proposed	$B?^(B $B#4(B $B$K(B $B<($9(B $B$h$&(B $B$K(B $B!"(B $B%P%k%V(B $B%*!<%P%i%C%W(B $BNL(B $B$O(B $B!"(B $B5[5$(B $BJ[(B $B#3(B $B#0(B $B$N(B $BJ[(B $B$N(B $B3QEY(B $B$,(B $B8:>/(B $B$7(B $B$F(B $B$$$k(B $B!#(B
	Baseline	$B?^(B $B#4(B $B$K(B $B<($9(B $B$h$&(B $B$K(B $B!"(B $B%P%k%V(B $B%*!<%P!<%i%C%W(B $BNL(B $B$,(B $B8:>/(B $B$7(B $B!"(B $B5[5$(B $BJ[(B $B#3(B $B$N(B $B:nMQ(B $B3Q(B $B$,(B $B8:>/(B $B$7(B $B$F(B $B$$$k(B $B!#(B
	RBMT	$B?^(B $B#4(B $B$K(B $B<($5(B $B$l$k(B $B$h$&(B $B$K(B $B!"(B $B%P%k%V(B $B!&(B $B%*!<%P!<%i%C%W(B $BNL(B $B$O(B $B!"(B $B5[$$9~$_(B $BJ[(B $B#3(B $B#0(B $B8:>/(B $B$N(B $B%P%k%V(B $B$r(B $BF0$+$9(B $B3QEY(B $B$K$D$l$F(B $B8:>/(B $B$9$k(B $B!#(B
	REFERENCE	$BF1(B $B?^(B $B#4(B $B$K(B $B<($9(B $B$h$&(B $B$K(B $B!"(B $B5[5$(B $B%P%k%V(B $B#3(B $B#0(B $B$N(B $B%P%k%V(B $B:nF0(B $B3Q(B $B$,(B $B>.$5$$(B $B$H$-(B $B$[$I(B $B!"(B $B%P%k%V(B $B%*!<%P!<%i%C%W(B $BNL(B $B$O(B $B>.$5$$(B
2	Input	Thereby, the retention mechanism 8 can be easily detached from the motherboard 2.
	Proposed	$B$3$l(B $B$K(B $B$h$j(B $B!"(B $BJ];}(B $B5!9=(B $B#8(B $B$O(B $B!"(B $B%^%6!<(B $B%\!<%I(B $B#2(B $B$+$i(B $BMF0W(B $B$K(B $BJ,N%(B $B$9$k(B $B$3$H(B $B$,(B $B$G$-$k(B $B!#(B
	Baseline	$B$3$l(B $B$K(B $B$h$j(B $B!"(B $BJ];}(B $B5!9=(B $B$O(B $B!"(B $B%^%6!<(B $B%\!<%I(B $B#2(B $B#8(B $B$KBP$7$F(B $BCeC&(B $B<+:_(B $B$H(B $B$5(B $B$l(B $B$F(B $B$$$k(B $B!#(B
	RBMT	$B$=$N(B $B$?$a(B $B$K(B $B!"(B $BJ];}(B $B%a%+%K%:%`(B $B#8(B $B$O(B $B!"(B $BMF0W(B $B$K(B $B%^%6!<(B $B%\!<%I(B $B#2(B $B$+$i(B $BJ,N%(B $B$9$k(B $B$3$H(B $B$,(B $B$G$-$k(B $B!#(B
	REFERENCE	$B$=$l(B $B$K(B $B$h$j(B $B!"(B $B%j(B $B%F%s%7%g%s(B $B%a%+%K%:%`(B $B#8(B $B$r(B $B%^%6!<(B $B%\!<%I(B $B#2(B $B$+$i(B $BMF0W(B $B$K(B $B
4	Input	FIG. 26 is a partially enlarged perspective view showing guides 250 of the disc array storage 200.
	Proposed	$B?^(B $B#2(B $B#6(B $B$O(B $B!"(B $B%G%#%9%/(B $B%"%l%$(B $B5-21(B $BAuCV(B $B#2(B $B#0(B $B#0(B $B$N(B $B%,%$%I(B $BIt(B $B#2(B $B#5(B $B#0(B $B$r(B $BItJ,(B $BE*(B $B$K(B $B3HBg(B $B$7(B $B$?(B $B
	Baseline	$B?^(B $B#2(B $B#6(B $B$O(B $B!"(B $B%,%$%I(B $B#2(B $B#5(B $B$N(B $B%G%#%9%/(B $B%"%l%$(B $BAuCV(B $B#2(B $B#0(B $B$N(B $B0lIt(B $B3HBg(B $B
	RBMT	$B?^(B $B#2(B $B#6(B $B$O(B $B!"(B $B%G%#%9%/(B $B!&(B $B%"%l%$(B $B5-21(B $BAuCV(B $B#2(B $B#0(B $B#0(B $B$N(B $B%,%$%I(B $B#2(B $B#5(B $B#0(B $B$r(B $B<($9(B $B!"(B $BItJ,(B $BE*(B $B$K(B $B3HBg(B $B$7(B $B$?(B $BF);k(B $B?^(B $B$G(B $B$"$k(B $B!#(B
	REFERENCE	$B$3$3(B $B$G(B $B!"(B $B?^(B $B#2(B $B#6(B $B$O(B $B!"(B $B%G%#%9%/(B $B%"%l%$(B $BAuCV(B $B#2(B $B#0(B $B#0(B $B$N(B $B%,%$%I(B $B#2(B $B#5(B $B#0(B $B$r(B $B@bL@(B $B$9$k(B $B$?$a(B $B$N(B $BItJ,(B $B3HBg(B $B
19	Input	When the sheathed wire is press-fitted into the slot portion 22 of each press-contacting piece portion 23 of the terminal 12 from the upper side, the insulating sheath of the sheathed wire is cut by the blade 21 of the slot portion 22, so that the internal conductor of the sheathed wire contacts the press-contacting piece portion 23.
	Proposed	$BHoJ$(B $BEE@~(B $B$r(B $B!"(B $B>eB&(B $B$+$i(B $BC<;R(B $B#1(B $B#2(B $B$N(B $B3F(B $B05(B $B@\(B $BJR(B $B#2(B $B#3(B $B$N(B $B%9%m%C%H(B $B#2(B $B#2(B $B$K(B $B05F~(B $B$9$k(B $B:](B $B!"(B $BHoJ$(B $BEE@~(B $B$N(B $B@d1o(B $BHoJ$(B $B$O(B $B%9%m%C%H(B $BIt(B $B#2(B $B#2(B $B$N(B $B%V%l!<%I(B $B#2(B $B#1(B $B$G(B $B@ZCG(B $B$5(B $B$l(B $B!"(B $BHoJ$(B $BEE@~(B $B$N(B $BFbIt(B $BF3(B $BEE(B $BBN(B $B$O(B $B!"(B $B05(B $B@\(B $BJR(B $B#2(B $B#3(B $B$K(B $BEv(B $B@\(B $B$9$k(B $B!#(B
	Baseline	$B$^$?(B $B!"(B $B05(B $B@\(B $BC<;R(B $B#1(B $B$N(B $B3F(B $B05(B $B@\(B $B?O(B $B#2(B $B#1(B $B$N(B $B3+8}(B $BIt(B $B#2(B $B#2(B $B$K(B $B@\?((B $B$7(B $B$F(B $BHoJ$(B $BEE@~(B $B$N(B $BHoJ$(B $BEE@~(B $B$N(B $BHoJ$(B $BIt(B $B$r(B $B@ZCG(B $B$7(B $B!"(B $B@d1o(B $BHoJ$(B $BEE@~(B $B$r(B $B05F~(B $B$9$k(B $B9BIt(B $B#2(B $B#2(B $B$N(B $BJR(B $B#2(B $B#3(B $B#2(B $B$N(B $B>eLL(B $B$+$i(B $BFbIt(B $BF3BN(B $BJR(B $B#2(B $B#3(B $B$G(B $B$"$k(B $B!#(B
	RBMT	$B>d(B $B$K(B $BG<$a(B $B$i$l(B $B$?(B $B%o%$%d!<(B $B$,(B $B>eIt(B $B$N(B $BB&(B $B$+$i(B $B$N(B $B%?!<%_%J%k(B $B#1(B $B#2(B $B$N(B $B3F(B $B05Gw(B $B$H(B $B@\?((B $B$9$k(B $BItJ,(B $BItJ,(B $B#2(B $B#3(B $B$N(B $B%9%m%C%H(B $BItJ,(B $B#2(B $B#2(B $B$K(B $B05Gw(B $BF~$l(B $B$i$l$k(B $B>l9g(B $B!"(B $B>d(B $B$K(B $BG<$a(B $B$i$l(B $B$?(B $B%o%$%d!<(B $B$N(B $B@d1o(B $B$9$k(B $B$5$d(B $B$O(B $B%9%m%C%H(B $BItJ,(B $B#2(B $B#2(B $B$N(B $BMU(B $B#2(B $B#1(B $B$K$h$C$F(B $B%+%C%H(B $B$5(B $B$l$k(B $B!#(B $B$=$N(B $B7k2L(B $B!"(B $B>d(B $B$K(B $BG<$a(B $B$i$l(B $B$?(B $B%o%$%d!<(B $B$N(B $BFbIt(B $BEAF3(B $BBN(B $B$O(B $B05Gw(B $B$H(B $B@\?((B $B$9$k(B $BItJ,(B $BItJ,(B $B#2(B $B#3(B $B$H(B $B@\?((B $B$9$k(B $B!#(B
	REFERENCE	$B$=$7$F(B $B!"(B $BC<;R(B $B#1(B $B#2(B $B$N(B $B>eJ}(B $B$+$i(B $B05(B $B@\(B $BJR(B $B#2(B $B#3(B $B$N(B $B%9%m%C%H(B $B#2(B $B#2(B $B$K(B $BHoJ$(B $BEE@~(B $B$,(B $B05F~(B $B$5(B $B$l$k(B $B$H(B $B!"(B $BHoJ$(B $BEE@~(B $B$N(B $B@d1o(B $BHoJ$(B $B$,(B $B%9%m%C%H(B $B#2(B $B#2(B $B$N(B $B?O(B $BIt(B $B#2(B $B#1(B $B$K$h$j(B $B@Z(B $BNv(B $B$5(B $B$l(B $B$F(B $BHoJ$(B $BEE@~(B $B$N(B $BFbIt(B $B$N(B $BF3BN(B $B$,(B $B05(B $B@\(B $BJR(B $B#2(B $B#3(B $B$K(B $B@\?((B $B$9$k(B $B!#(B

``Input'' : input sentence.
``Proposed'' : output of proposed method with parameter optimizing.
``Baseline'' : output of moses with no parameter tuning.
``RBMT'' : output of rule based machine trainslation.
``REFERENCE'' : reference (correct) sentence .

$B $B$3$NJ8=q$K$D$$$F(B... $B>e$X(B: Statistical Machine Translation with $BLa$k(B: Acknowledgements