検索・分析のための手順文章からの意味構造抽出
Contents
1.2.
3.
4.
5.
6.
7.
8.
9. U6 1HHHHHHHHHHHHH
10. 3 2 6 At Y mn Ac d obj J ES Af We a Ac F 3 2 7 Sf T Ac Af OO FO T 00 Sf ES Q op 99 yo A O00 SfH D Af SE T HH A D SfD Af 3 3 2 T comp F comp SEO O0 D F T he E 3 2 8 St Ac Af T comp TT Ac HH StH TU HH Ac a TH StO OO Ac FH JD Ac 3 3 F O Ac 3 3 3 F eq F part of S F set
11. 2 DUDU
12. DO UU HH F HH HHHHHHHHHHH T UU HHHHHHHH D 0000 HHHHHHH Q OD HHHHHHH Ac 000000 0000 Af 00000 HHHH st HHHHH st 00000 3 HHHHHHHHH 3 4 3 1 3 1 2 3 2 3 DAG Directed Acyclic Graph DAG 3 2 1 3 2 1 F D O 0 1 HHHHHHHHHHHHHHHHHHHHH Mori Momouchi 9 Hamada 10 Hamada 2 1 2 11 6 7 12 MST Maximum Spanning Tree MST McDonald 6 13 MST 14 McDonald 13 2 McDonald
13. 3 HHHHHHHHHHH 1000000000 UU OO IHHHHHHHH 000 000000 HHH 000 000 200000 6 51 36 34 13 91 22 43 118 65 180 51 HH 66HHH 6 91 48 05 22 12 25 92 132 95 198 70 H Ge 4 HHHHHHHHH T part of n T SS K sse GE ori 8 3 3 5 A eq 2 DAG U s 4 us 2 MST MST A eq 1 MST gu d obj e dd 000 OD 2 Ac 2 MST J d obj 00 a F s Ac 000 BD 2 Ac E D s Ac py 4 1 MST MST D Ge V n va os tv V G 2 Vi Lon 00 40 59 mu vat S srv 3 3 6 V tm Bi Ui oo II OO ID OO MST V tm Iu 1AT O Ac G argmax K Score vi vj Geg le a 3 3 7 other mod vis E Chu Liu Edmonds 15 16 Sf other mod Ac Score vi vj other mod 45 D 0 Ac unio S exp O feat vi v 1 3 4 e 5 exp O feat vi w w VNt vil 3 Mor 3 Cook pad D UP V2 VxvV o feat vi vj Vi Uj oo http cookpad com 2014 27 1000000
14. 1i VHHHHHH 0 3 2 4 HHHHHH GHHHHHHHHHH 025 ES 3 AQ SeoreHHHHHHHHHHHHHHH 0 2 4 n 0 0 15 5 for v w A do a 0 1 6 if Score v w gt Penalty n then 0 05 T G Gc v w 0 8 nem l 01234567 8 9101112131415161718 9 end if 5 m 0 HHHHHHHHHHHHHHHHH 11 return G 0 4 DAGHHHHHHHHH 17 5 o Vi Ve melu D T Vi vi We MST E 4 3 KS Va exp O feat vi we C lei 5 exp O feat vi w 2 weVvi ve SVM DAG 1 2 V D w 2 D 4 4 V v w MST V v w 2 2 Vi Uj 4 2 DAG DAG 4 MST e j i 9 i Uj 4 Score MST e vi vj Penalty n n e vill vj e vi vj 3 MST vi Uj A Vi Uj A Vi Uj 9 vi Uj Vi Uj A Vi j Penalty n is Ae o vi A Uj Vi Uj A A Ui e vvi Uj A Vi Vj DAG 9 vi Gi Gi Avill vj 3 A 5 D4 HHHHHHHHHHHHHH
15. D 1 2 KABUTY 2 1 3 4 5 D 3 X BEBE T F part of 1 d obj F eq F part of
16. 2 3 1 0000000
17. HH HHH HH HH HH 000 subj 84 2 309 367 93 5 58 62 85 5 367 429 d obj 90 4 2684 2970 54 3 89 164 88 5 2773 3134 i obj 72 9 902 1238 45 1 93 206 68 9 995 1440 F comp 78 1 100 128 0 0 0 1 77 5 100 129 T comp 78 5 205 261 0 0 0 2 77 9 205 263 84 6 4200 4964 55 2 240 435 80 0 4440 5399 F eq 1 9 111 23 2 102 519 17 6 111 630 F part of ari 125 300 29 3 51 174 37 1 176 474 F set 0 0 13 59 1 17 3 3 1 30 T eq 33 3 7 21 4 5 3 66 11 5 10 87 T part of 3 2 47 6 5 2 31 5 1 4 78 A eq 0 0 6 10 1 99 10 1 105 28 7 143 498 17 7 160 906 21 6 303 1404 V tm 49 5 104 210 0 0 0 2 49 1 104 212 other mod 56 0 385 687 4 8 1 21 54 5 386 708 54 5 489 897 4 3 1 23 53 3 490 920 76 0 4832 6359 27 3 401 1364 67 8 5233 7723 of the Association for Computational Linguistics pp 99 DAG DAG 106 2005 8 Shinsuke Mori Tetsuro Sasada Yoko Yamakata and Koichiro Yoshino A machine learning approach to recipe DAG text processing In Proceedings of Cooking with Computer MST workshop 2012 9 Yoshio Momouchi Control s
18. UU 000 000 FO HHHHH 0 000000 HHHHH 759 69 4 72 5 subj 2 15 HHHH HHHH 80 9 75 8 78 3 d obj 15 67 0000 00000 724 67 8 70 0 i obj 7 22 F comp 0 65 T comp 1 32 5 F eq 3 15 F part of 2 37 DAG F set 0 15 T eq 0 44 T part of 0 39 5 1 A eq 0 53 3 V tm 1 06 other mod 3 54 38 62 10 Hamada 10 Hamada 5 DAG F 5 subj d obj i obj i obj Nsys Nref Nint F subj N subj _ int i Nsys Nint Nref F eq T F part of A eq F 2Nint Nref Nsys 5 2 6 1 4 Hamada 10 6 1 F eq T F part of A eq 1 F eq 18 180 6 0 0 0 F 10 90 1 05 HHHHHHHHHHHHHHHHHHHHHHHHH
19. 3 0200000000000000000 02 00000 HHHHHHTHHHH HH D subj HH 00 D d 0 0 000 TO AcH HH T 3 2 3 D 0000 SH Sen HHHHHHHHHHHHHHHHHHHHHHHHHHH ii HHHHHHHHHHHHHHHHHHHHHH UR 0 000 00 0 00 nn Acn HH F part of O00000 F set HHHHH HHH D T eq 00000 T part of 00000 swa wp H ro HHHHH 000000000000000000000000000 V tm 00000 HHHHHHHHHHHHHHHHHHHH other mod 00000000 0 00 F 0 0000 000 0 0000000 FO00000000000000000 3 2 5 AcHHHHHRE on 000000000000000000000000000 HHHHHHHHHHHHHHHHHHHHHHHHHHHH HHH D FD D Ael O HHHHHHHHHHHHHHHHHHH pA Tet T D Af 0 HHHH FH HHHHHHHHHHHHHHHHHHHHHHHHHH i 0000000000000000 0000000000 H HHHH AcHH
20. JD 2 i F eq e sa F e 2 mie 70 F 3 3 1 subj d obj i obj AR F part of F part of 00 D OO F Af m ve PS mon e D O FO OO AA subj D F gt Af m F set 00 00M Af F D O00 FO O0 Ac mnn F Es jon r F set obj ELES F mun me 9 mna ae 3 3 4 T eq T part of T Ac D OO FO HH Ac
21. DEIM Forum 2014 C3 2 HHHHHHHHHHHHUHHHHHHHHHLH JUI JH do H P HH THHHHHHHHHH O 606 8501HHHHHHHHHH ffHHHHHHHHHHHHHHHH O 606 8501 000000000 E mail tfforestOi kyoto u ac jp D Du 1
22. tructures for actions in proce dural texts and pt chart In Proceedings of the Eighth In ternational Conference on Computational Linguistics pp 108 114 1980 10 Reiko Hamada Ichiro Ide Shuichi Sakai and Hidehiko Tanaka Structural analysis of cooking preparation steps in japanese In Proceedings of the fifth international work shop on Information retrieval with Asian languages No 8 in IRAL 00 pp 157 164 2000 11 Ryan McDonald and Fernando Pereira Online learning of approximate dependency parsing algorithms In Proceed ings of the Eleventh European Chapter of the Association for Computational Linguistics pp 81 88 2006 1 Liping Wang Qing Li Na Li Guozhu Dong and Yu Yang 12 Kenji Sagae and Alon Lavie A best first probabilistic shift Substructure similarity measurement in chinese recipes In reduce parser In Proceedings of the 21th International Con Proceedings of the 17th international conference on World ference on Computational Linguistics 2006 Wide Web pp 978 988 2008 13 Ryan McDonald and Joakim Nivre Analyzing and integrat 2 0000 000 0000 000000000000000 ing dependency parsers Computational Linguistics Vol 37 HHHHHHHHHHHHHHHHHHHH HHHHHHHH No 4 pp 197 230 2011 Vol J90 DII No 10 pp 2817 2829 2007 14 Daniel Flannery Yusuke Miyao Graham Neubig and Shin 3 Shinsuke Mori Hirokuni Maeta Yoko Yamakata and Tet suke Mori A pointwise approach to training dependency s
23. uro Sasada Flow graph corpus from recipe texts In Pro parsers from partially annotated corpora Journal of Natu ceedings of the Nineth International Conference on Lan ral Language Processing Vol 19 No 3 2012 guage Resources and Evaluation 2014 15 Yoeng Jin Chu and Tseng Hong Liu On the shortest ar 4 000 0000 0000 0000 0000000000 borescence of a directed graph Science Sinica Vol 14 pp HHHHHHHHHH HHHHHHHHHH H NL214 1396 1400 1965 2013 16 Jack Edmonds Optimum branchings Journal Research of 5 0000 000 0000 0000 0000 00 0 0 the National Bureau of Standards Vol 71B pp 233 240 HHHHHHHHHHHHHHHHHHHHH HHHHHHH 1967 Vol J86 DII No 11 pp 1647 1656 2003 17 Vincent J Della Pietra Adam L Berger Stephen A 6 Ryan McDonald Fernando Pereira Kiril Ribarov and Jan Della Pietra A maximum entropy approach to natural lan Hajic Non projective dependency parsing using spanning guage processing Computational Linguistics Vol 22 No 1 tree algorithms In Proceedings of Human Language Tech 1996 nology Conference and Conference on Empirical Methods as 0000 0000 00 0 0000 0000 00000 in Natural Language Processing pp 523 530 2005 00000000000000000000000 00000 7 Joakim Nivre and Jens Nilsson Pseudo projective depen OOOCGCCoOOgO 2014 dency parsing In Proceedings of the 43rd Annual Meeting
Download Pdf Manuals
Related Search