1: %!PS-Adobe-3.0

2: %%Title: (ACL'00 FIN adjusted)

3: %%Creator: (Microsoft Word 5.1: LaserWriter 8 N1-8.2.2)

4: %%CreationDate: (17:35 vrijdag, 22 september 2000)

5: %%For: (rens)

6: %%Pages: 8

7: %%DocumentFonts: Times-Bold Symbol Times-Roman Times-Italic Times-BoldItalic

8: %%DocumentNeededFonts: Times-Bold Symbol Times-Roman Times-Italic Times-BoldItalic

9: %%DocumentSuppliedFonts:

10: %%DocumentData: Clean7Bit

11: %%PageOrder: Ascend

12: %%Orientation: Portrait

13: %%DocumentMedia: Default 595 842 0 () ()

14: %ADO_ImageableArea: 29 31 567 812

15: %%EndComments

16: userdict begin/dscInfo 5 dict dup begin

17: /Title(ACL'00 FIN adjusted)def

18: /Creator(Microsoft Word 5.1: LaserWriter 8 N1-8.2.2)def

19: /CreationDate(17:35 vrijdag, 22 september 2000)def

20: /For(rens)def

21: /Pages 1 def

22: end def end

23: /md 198 dict def md begin/currentpacking where {pop /sc_oldpacking currentpacking def true setpacking}if

24: %%BeginFile: adobe_psp_basic

25: %%Copyright: Copyright 1990-1993 Adobe Systems Incorporated. All Rights Reserved.

26: /bd{bind def}bind def

27: /xdf{exch def}bd

28: /xs{exch store}bd

29: /ld{load def}bd

30: /Z{0 def}bd

31: /T/true

32: /F/false

33: /:L/lineto

34: /lw/setlinewidth

35: /:M/moveto

36: /rl/rlineto

37: /rm/rmoveto

38: /:C/curveto

39: /:T/translate

40: /:K/closepath

41: /:mf/makefont

42: /gS/gsave

43: /gR/grestore

44: /np/newpath

45: 14{ld}repeat

46: /$m matrix def

47: /av 81 def

48: /por true def

49: /normland false def

50: /psb-nosave{}bd

51: /pse-nosave{}bd

52: /us Z

53: /psb{/us save store}bd

54: /pse{us restore}bd

55: /level2

56: /languagelevel where

57: {

58: pop languagelevel 2 ge

59: }{

60: false

61: }ifelse

62: def

63: /featurecleanup

64: {

65: stopped

66: cleartomark

67: countdictstack exch sub dup 0 gt

68: {

69: {end}repeat

70: }{

71: pop

72: }ifelse

73: }bd

74: /noload Z

75: /startnoload

76: {

77: {/noload save store}if

78: }bd

79: /endnoload

80: {

81: {noload restore}if

82: }bd

83: level2 startnoload

84: /setjob

85: {

86: statusdict/jobname 3 -1 roll put

87: }bd

88: /setcopies

89: {

90: userdict/#copies 3 -1 roll put

91: }bd

92: level2 endnoload level2 not startnoload

93: /setjob

94: {

95: 1 dict begin/JobName xdf currentdict end setuserparams

96: }bd

97: /setcopies

98: {

99: 1 dict begin/NumCopies xdf currentdict end setpagedevice

100: }bd

101: level2 not endnoload

102: /pm Z

103: /mT Z

104: /sD Z

105: /realshowpage Z

106: /initializepage

107: {

108: /pm save store mT concat

109: }bd

110: /endp

111: {

112: pm restore showpage

113: }def

114: /$c/DeviceRGB def

115: /rectclip where

116: {

117: pop/rC/rectclip ld

118: }{

119: /rC

120: {

121: np 4 2 roll

122: :M

123: 1 index 0 rl

124: 0 exch rl

125: neg 0 rl

126: :K

127: clip np

128: }bd

129: }ifelse

130: /rectfill where

131: {

132: pop/rF/rectfill ld

133: }{

134: /rF

135: {

136: gS

137: np

138: 4 2 roll

139: :M

140: 1 index 0 rl

141: 0 exch rl

142: neg 0 rl

143: fill

144: gR

145: }bd

146: }ifelse

147: /rectstroke where

148: {

149: pop/rS/rectstroke ld

150: }{

151: /rS

152: {

153: gS

154: np

155: 4 2 roll

156: :M

157: 1 index 0 rl

158: 0 exch rl

159: neg 0 rl

160: :K

161: stroke

162: gR

163: }bd

164: }ifelse

165: %%EndFile

166: %%BeginFile: adobe_psp_colorspace_level1

167: %%Copyright: Copyright 1991-1993 Adobe Systems Incorporated. All Rights Reserved.

168: /G/setgray ld

169: /:F/setrgbcolor ld

170: %%EndFile

171: %%BeginFile: adobe_psp_uniform_graphics

172: %%Copyright: Copyright 1990-1993 Adobe Systems Incorporated. All Rights Reserved.

173: /@a

174: {

175: np :M 0 rl :L 0 exch rl 0 rl :L fill

176: }bd

177: /@b

178: {

179: np :M 0 rl 0 exch rl :L 0 rl 0 exch rl fill

180: }bd

181: /arct where

182: {

183: pop

184: }{

185: /arct

186: {

187: arcto pop pop pop pop

188: }bd

189: }ifelse

190: /x1 Z

191: /x2 Z

192: /y1 Z

193: /y2 Z

194: /rad Z

195: /@q

196: {

197: /rad xs

198: /y2 xs

199: /x2 xs

200: /y1 xs

201: /x1 xs

202: np

203: x2 x1 add 2 div y1 :M

204: x2 y1 x2 y2 rad arct

205: x2 y2 x1 y2 rad arct

206: x1 y2 x1 y1 rad arct

207: x1 y1 x2 y1 rad arct

208: fill

209: }bd

210: /@s

211: {

212: /rad xs

213: /y2 xs

214: /x2 xs

215: /y1 xs

216: /x1 xs

217: np

218: x2 x1 add 2 div y1 :M

219: x2 y1 x2 y2 rad arct

220: x2 y2 x1 y2 rad arct

221: x1 y2 x1 y1 rad arct

222: x1 y1 x2 y1 rad arct

223: :K

224: stroke

225: }bd

226: /@i

227: {

228: np 0 360 arc fill

229: }bd

230: /@j

231: {

232: gS

233: np

234: :T

235: scale

236: 0 0 .5 0 360 arc

237: fill

238: gR

239: }bd

240: /@e

241: {

242: np

243: 0 360 arc

244: :K

245: stroke

246: }bd

247: /@f

248: {

249: np

250: $m currentmatrix

251: pop

252: :T

253: scale

254: 0 0 .5 0 360 arc

255: :K

256: $m setmatrix

257: stroke

258: }bd

259: /@k

260: {

261: gS

262: np

263: :T

264: 0 0 :M

265: 0 0 5 2 roll

266: arc fill

267: gR

268: }bd

269: /@l

270: {

271: gS

272: np

273: :T

274: 0 0 :M

275: scale

276: 0 0 .5 5 -2 roll arc

277: fill

278: gR

279: }bd

280: /@m

281: {

282: np

283: arc

284: stroke

285: }bd

286: /@n

287: {

288: np

289: $m currentmatrix

290: pop

291: :T

292: scale

293: 0 0 .5 5 -2 roll arc

294: $m setmatrix

295: stroke

296: }bd

297: %%EndFile

298: %%BeginFile: adobe_psp_basic_text

299: %%Copyright: Copyright 1990-1993 Adobe Systems Incorporated. All Rights Reserved.

300: /S/show ld

301: /A{

302: 0.0 exch ashow

303: }bd

304: /R{

305: 0.0 exch 32 exch widthshow

306: }bd

307: /W{

308: 0.0 3 1 roll widthshow

309: }bd

310: /J{

311: 0.0 32 4 2 roll 0.0 exch awidthshow

312: }bd

313: /V{

314: 0.0 4 1 roll 0.0 exch awidthshow

315: }bd

316: /fcflg true def

317: /fc{

318: fcflg{

319: vmstatus exch sub 50000 lt{

320: (%%[ Warning: Running out of memory ]%%\r)print flush/fcflg false store

321: }if pop

322: }if

323: }bd

324: /$f[1 0 0 -1 0 0]def

325: /:ff{$f :mf}bd

326: /MacEncoding StandardEncoding 256 array copy def

327: MacEncoding 39/quotesingle put

328: MacEncoding 96/grave put

329: /Adieresis/Aring/Ccedilla/Eacute/Ntilde/Odieresis/Udieresis/aacute

330: /agrave/acircumflex/adieresis/atilde/aring/ccedilla/eacute/egrave

331: /ecircumflex/edieresis/iacute/igrave/icircumflex/idieresis/ntilde/oacute

332: /ograve/ocircumflex/odieresis/otilde/uacute/ugrave/ucircumflex/udieresis

333: /dagger/degree/cent/sterling/section/bullet/paragraph/germandbls

334: /registered/copyright/trademark/acute/dieresis/notequal/AE/Oslash

335: /infinity/plusminus/lessequal/greaterequal/yen/mu/partialdiff/summation

336: /product/pi/integral/ordfeminine/ordmasculine/Omega/ae/oslash

337: /questiondown/exclamdown/logicalnot/radical/florin/approxequal/Delta/guillemotleft

338: /guillemotright/ellipsis/space/Agrave/Atilde/Otilde/OE/oe

339: /endash/emdash/quotedblleft/quotedblright/quoteleft/quoteright/divide/lozenge

340: /ydieresis/Ydieresis/fraction/currency/guilsinglleft/guilsinglright/fi/fl

341: /daggerdbl/periodcentered/quotesinglbase/quotedblbase/perthousand

342: /Acircumflex/Ecircumflex/Aacute/Edieresis/Egrave/Iacute/Icircumflex/Idieresis/Igrave

343: /Oacute/Ocircumflex/apple/Ograve/Uacute/Ucircumflex/Ugrave/dotlessi/circumflex/tilde

344: /macron/breve/dotaccent/ring/cedilla/hungarumlaut/ogonek/caron

345: MacEncoding 128 128 getinterval astore pop

346: level2 startnoload

347: /copyfontdict

348: {

349: findfont dup length dict

350: begin

351: {

352: 1 index/FID ne{def}{pop pop}ifelse

353: }forall

354: }bd

355: level2 endnoload level2 not startnoload

356: /copyfontdict

357: {

358: findfont dup length dict

359: copy

360: begin

361: }bd

362: level2 not endnoload

363: md/fontname known not{

364: /fontname/customfont def

365: }if

366: /Encoding Z

367: /:mre

368: {

369: copyfontdict

370: /Encoding MacEncoding def

371: fontname currentdict

372: end

373: definefont :ff def

374: }bd

375: /:bsr

376: {

377: copyfontdict

378: /Encoding Encoding 256 array copy def

379: Encoding dup

380: }bd

381: /pd{put dup}bd

382: /:esr

383: {

384: pop pop

385: fontname currentdict

386: end

387: definefont :ff def

388: }bd

389: /scf

390: {

391: scalefont def

392: }bd

393: /scf-non

394: {

395: $m scale :mf setfont

396: }bd

397: /ps Z

398: /fz{/ps xs}bd

399: /sf/setfont ld

400: /cF/currentfont ld

401: /mbf

402: {

403: /makeblendedfont where

404: {

405: pop

406: makeblendedfont

407: /ABlend exch definefont

408: }{

409: pop

410: }ifelse

411: def

412: }def

413: %%EndFile

414: %%BeginFile: adobe_psp_derived_styles

415: %%Copyright: Copyright 1990-1993 Adobe Systems Incorporated. All Rights Reserved.

416: /wi

417: version(23.0)eq

418: {

419: {

420: gS 0 0 0 0 rC stringwidth gR

421: }bind

422: }{

423: /stringwidth load

424: }ifelse

425: def

426: /$o 1. def

427: /gl{$o G}bd

428: /ms{:M S}bd

429: /condensedmtx[.82 0 0 1 0 0]def

430: /:mc

431: {

432: condensedmtx :mf def

433: }bd

434: /extendedmtx[1.18 0 0 1 0 0]def

435: /:me

436: {

437: extendedmtx :mf def

438: }bd

439: /basefont Z

440: /basefonto Z

441: /dxa Z

442: /dxb Z

443: /dxc Z

444: /dxd Z

445: /dsdx2 Z

446: /bfproc Z

447: /:fbase

448: {

449: dup/FontType get 0 eq{

450: dup length dict begin

451: dup{1 index/FID ne 2 index/UniqueID ne and{def}{pop pop}ifelse}forall

452: /FDepVector exch/FDepVector get[exch/:fbase load forall]def

453: }/bfproc load ifelse

454: /customfont currentdict end definefont

455: }bd

456: /:mo

457: {

458: /bfproc{

459: dup dup length 2 add dict

460: begin

461: {

462: 1 index/FID ne 2 index/UniqueID ne and{def}{pop pop}ifelse

463: }forall

464: /PaintType 2 def

465: /StrokeWidth .012 0 FontMatrix idtransform pop def

466: /customfont currentdict

467: end

468: definefont

469: 8 dict begin

470: /basefonto xdf

471: /basefont xdf

472: /FontType 3 def

473: /FontMatrix[1 0 0 1 0 0]def

474: /FontBBox[0 0 1 1]def

475: /Encoding StandardEncoding def

476: /BuildChar

477: {

478: exch begin

479: basefont setfont

480: ( )dup 0 4 -1 roll put

481: dup wi

482: setcharwidth

483: 0 0 :M

484: gS

485: gl

486: dup show

487: gR

488: basefonto setfont

489: show

490: end

491: }def

492: }store :fbase

493: }bd

494: /:mso

495: {

496: /bfproc{

497: 7 dict begin

498: /basefont xdf

499: /FontType 3 def

500: /FontMatrix[1 0 0 1 0 0]def

501: /FontBBox[0 0 1 1]def

502: /Encoding StandardEncoding def

503: /BuildChar

504: {

505: exch begin

506: sD begin

507: /dxa 1 ps div def

508: basefont setfont

509: ( )dup 0 4 -1 roll put

510: dup wi

511: 1 index 0 ne

512: {

513: exch dxa add exch

514: }if

515: setcharwidth

516: dup 0 0 ms

517: dup dxa 0 ms

518: dup dxa dxa ms

519: dup 0 dxa ms

520: gl

521: dxa 2. div dup ms

522: end

523: end

524: }def

525: }store :fbase

526: }bd

527: /:ms

528: {

529: /bfproc{

530: dup dup length 2 add dict

531: begin

532: {

533: 1 index/FID ne 2 index/UniqueID ne and{def}{pop pop}ifelse

534: }forall

535: /PaintType 2 def

536: /StrokeWidth .012 0 FontMatrix idtransform pop def

537: /customfont currentdict

538: end

539: definefont

540: 8 dict begin

541: /basefonto xdf

542: /basefont xdf

543: /FontType 3 def

544: /FontMatrix[1 0 0 1 0 0]def

545: /FontBBox[0 0 1 1]def

546: /Encoding StandardEncoding def

547: /BuildChar

548: {

549: exch begin

550: sD begin

551: /dxb .05 def

552: basefont setfont

553: ( )dup 0 4 -1 roll put

554: dup wi

555: exch dup 0 ne

556: {

557: dxb add

558: }if

559: exch setcharwidth

560: dup dxb .01 add 0 ms

561: 0 dxb :T

562: gS

563: gl

564: dup 0 0 ms

565: gR

566: basefonto setfont

567: 0 0 ms

568: end

569: end

570: }def

571: }store :fbase

572: }bd

573: /:mss

574: {

575: /bfproc{

576: 7 dict begin

577: /basefont xdf

578: /FontType 3 def

579: /FontMatrix[1 0 0 1 0 0]def

580: /FontBBox[0 0 1 1]def

581: /Encoding StandardEncoding def

582: /BuildChar

583: {

584: exch begin

585: sD begin

586: /dxc 1 ps div def

587: /dsdx2 .05 dxc 2 div add def

588: basefont setfont

589: ( )dup 0 4 -1 roll put

590: dup wi

591: exch dup 0 ne

592: {

593: dsdx2 add

594: }if

595: exch setcharwidth

596: dup dsdx2 .01 add 0 ms

597: 0 .05 dxc 2 div sub :T

598: dup 0 0 ms

599: dup dxc 0 ms

600: dup dxc dxc ms

601: dup 0 dxc ms

602: gl

603: dxc 2 div dup ms

604: end

605: end

606: }def

607: }store :fbase

608: }bd

609: /:msb

610: {

611: /bfproc{

612: 7 dict begin

613: /basefont xdf

614: /FontType 3 def

615: /FontMatrix[1 0 0 1 0 0]def

616: /FontBBox[0 0 1 1]def

617: /Encoding StandardEncoding def

618: /BuildChar

619: {

620: exch begin

621: sD begin

622: /dxd .03 def

623: basefont setfont

624: ( )dup 0 4 -1 roll put

625: dup wi

626: 1 index 0 ne

627: {

628: exch dxd add exch

629: }if

630: setcharwidth

631: dup 0 0 ms

632: dup dxd 0 ms

633: dup dxd dxd ms

634: 0 dxd ms

635: end

636: end

637: }def

638: }store :fbase

639: }bd

640: /italicmtx[1 0 -.212557 1 0 0]def

641: /:mi

642: {

643: italicmtx :mf def

644: }bd

645: /:v

646: {

647: [exch dup/FontMatrix get exch

648: dup/FontInfo known

649: {

650: /FontInfo get

651: dup/UnderlinePosition known

652: {

653: dup/UnderlinePosition get

654: 2 index 0

655: 3 1 roll

656: transform

657: exch pop

658: }{

659: .1

660: }ifelse

661: 3 1 roll

662: dup/UnderlineThickness known

663: {

664: /UnderlineThickness get

665: exch 0 3 1 roll

666: transform

667: exch pop

668: abs

669: }{

670: pop pop .067

671: }ifelse

672: }{

673: pop pop .1 .067

674: }ifelse

675: ]

676: }bd

677: /$t Z

678: /$p Z

679: /$s Z

680: /:p

681: {

682: aload pop

683: 2 index mul/$t xs

684: 1 index mul/$p xs

685: .012 mul/$s xs

686: }bd

687: /:m

688: {gS

689: 0 $p rm

690: $t lw

691: 0 rl stroke

692: gR

693: }bd

694: /:n

695: {

696: gS

697: 0 $p rm

698: $t lw

699: 0 rl

700: gS

701: gl

702: stroke

703: gR

704: strokepath

705: $s lw

706: /setstrokeadjust where{pop

707: currentstrokeadjust true setstrokeadjust stroke setstrokeadjust

708: }{

709: stroke

710: }ifelse

711: gR

712: }bd

713: /:o

714: {gS

715: 0 $p rm

716: $t 2 div dup rm

717: $t lw

718: dup 0 rl

719: stroke

720: gR

721: :n

722: }bd

723: %%EndFile

724: /currentpacking where {pop sc_oldpacking setpacking}if end

725: %%EndProlog

726: %%BeginSetup

727: md begin

728: countdictstack[{

729: %%BeginFeature: *ManualFeed False

730: 1 dict dup /ManualFeed false put setpagedevice

731: %%EndFeature

732: }featurecleanup

733: countdictstack[{

734: %%BeginFeature: *InputSlot Cassette

735: 

736: %%EndFeature

737: }featurecleanup

738: countdictstack[{

739: %%BeginFeature: *PageRegion A4Small

740: 

741:     3 dict dup /PageSize [595 842] put dup /ImagingBBox [25 25 570 817] put

742: 	dup /InputAttributes 1 dict dup 0 1 dict dup

743: 	/PageSize [595 842] put put put setpagedevice

744: %%EndFeature

745: }featurecleanup

746: (rens)setjob

747: /mT[1 0 0 -1 29 812]def

748: /sD 16 dict def

749: 300 level2{1 dict dup/WaitTimeout 4 -1 roll put setuserparams}{statusdict/waittimeout 3 -1 roll put}ifelse

750: %%IncludeFont: Times-Bold

751: %%IncludeFont: Symbol

752: %%IncludeFont: Times-Roman

753: %%IncludeFont: Times-Italic

754: %%IncludeFont: Times-BoldItalic

755: /f0_1/Times-Bold

756: :mre

757: /f0_15 f0_1 15 scf

758: /f0_12 f0_1 12 scf

759: /f0_11 f0_1 11 scf

760: /f1_1/Symbol

761: :bsr

762: 240/apple pd

763: :esr

764: /f1_16 f1_1 16 scf

765: /f1_14 f1_1 14 scf

766: /f1_12 f1_1 12 scf

767: /f1_11 f1_1 11 scf

768: /f1_10 f1_1 10 scf

769: /f2_1 f1_1

770: def

771: /f2_12 f2_1 12 scf

772: /f2_11 f2_1 11 scf

773: /f3_1/Times-Roman

774: :mre

775: /f3_12 f3_1 12 scf

776: /f3_11 f3_1 11 scf

777: /f3_10 f3_1 10 scf

778: /f3_9 f3_1 9 scf

779: /f3_4 f3_1 4 scf

780: /f4_1/Times-Italic

781: :mre

782: /f4_12 f4_1 12 scf

783: /f4_11 f4_1 11 scf

784: /f4_10 f4_1 10 scf

785: /f4_9 f4_1 9 scf

786: /f4_4 f4_1 4 scf

787: /f5_1 f1_1

788: :mi

789: /f5_12 f5_1 12 scf

790: /f5_11 f5_1 11 scf

791: /f5_10 f5_1 10 scf

792: /f6_1/Times-BoldItalic

793: :mre

794: /f6_11 f6_1 11 scf

795: /f7_1 f2_1

796: :mi

797: /f7_11 f7_1 11 scf

798: /Courier findfont[10 0 0 -10 0 0]:mf setfont

799: %%EndSetup

800: %%Page: 1 1

801: %%BeginPageSetup

802: gsave %matrix defaultmatrix setmatrix

803: 90 rotate 4 72 mul .55 -72 mul moveto /Times-Roman findfont

804: 20 scalefont setfont 0.3 setgray (arXiv:cs.CL/0009026   27 Sep 2000) show grestore

805: initializepage

806: (rens; page: 1 of 8)setjob

807: %%EndPageSetup

808: gS 0 0 538 781 rC

809: 49 70 :M

810: f0_15 sf

811: .092 .009(An Improved Parser for Data-Oriented Lexical-Functional Analysis)J

812: 242 98 :M

813: f0_12 sf

814: .587 .059(Rens Bod)J

815: 87 113 :M

816: f3_12 sf

817: -.044(Informatics Research Institute, University of Leeds, Leeds LS2 9JT, UK, &)A

818: 92 128 :M

819: -.061(Institute for Logic, Language and Computation, University of Amsterdam)A

820: 217 143 :M

821: -.053(rens@scs.leeds.ac.uk)A

822: 127 180 :M

823: f0_12 sf

824: .051(Abstract)A

825: 41 194 :M

826: f3_11 sf

827: 2.506 .251(We present an LFG-DOP parser )J

828: 206 194 :M

829: 2.332 .233(which uses)J

830: 41 208 :M

831: -.049(fragments from LFG-annotated sentences to )A

832: 237 208 :M

833: -.301(parse)A

834: 41 222 :M

835: .372 .037(new sentences. Experiments with )J

836: 194 222 :M

837: .527 .053(the Verbmobil)J

838: 41 236 :M

839: .284 .028(and Homecentre corpora show that )J

840: 201 236 :M

841: .815 .081(\(1\) Viterbi )J

842: 254 236 :M

843: f4_11 sf

844: (n)S

845: 41 250 :M

846: f3_11 sf

847: .448 .045(best search performs about 100 times faster )J

848: 241 250 :M

849: -.313(than)A

850: 41 264 :M

851: .174 .017(Monte Carlo search while both achieve )J

852: 219 264 :M

853: .713 .071(the same)J

854: 41 278 :M

855: .47 .047(accuracy; \(2\) )J

856: 104 278 :M

857: 1.6 .16(the DOP hypothesis which states)J

858: 41 292 :M

859: -.14(that parse accuracy increases )A

860: 168 292 :M

861: -.031(with increasing frag-)A

862: 41 306 :M

863: .55 .055(ment size is confirmed for LFG-DOP; \(3\) )J

864: 234 306 :M

865: -.145(LFG-)A

866: 41 320 :M

867: 3.33 .333(DOP's )J

868: 78 320 :M

869: 2.149 .215(relative frequency estimator performs)J

870: 41 334 :M

871: -.023(worse than a discounted frequency estimator; and)A

872: 41 348 :M

873: 1.835 .184(\(4\) LFG-DOP significantly outperforms Tree)J

874: 256 348 :M

875: (-)S

876: 41 362 :M

877: -.072(DOP if evaluated on tree structures only.)A

878: 41 386 :M

879: f0_12 sf

880: .199 .02(1  Introduction)J

881: 41 400 :M

882: f3_11 sf

883: .27 .027(Data-Oriented Parsing \(DOP\) )J

884: 177 400 :M

885: .801 .08(models learn how)J

886: 41 414 :M

887: 3.03 .303(to provide linguistic representations )J

888: 226 414 :M

889: 4.698 .47(for an)J

890: 41 428 :M

891: -.112(unlimited set of )A

892: 112 428 :M

893: -.006(utterances by generalizing from a)A

894: 41 442 :M

895: 1.518 .152(given corpus of )J

896: 121 442 :M

897: .944 .094(properly annotated exemplars.)J

898: 41 456 :M

899: 5.209 .521(They operate by decomposing the )J

900: 236 456 :M

901: -.36(given)A

902: 41 470 :M

903: -.022(representations )A

904: 111 470 :M

905: .697 .07(into \(arbitrarily large\) fragments)J

906: 41 484 :M

907: 1.34 .134(and recomposing those )J

908: 154 484 :M

909: 1.761 .176(pieces to analyze new)J

910: 41 498 :M

911: -.136(utterances. A )A

912: 101 498 :M

913: .225 .023(probability model is used to choose)J

914: 41 512 :M

915: 2.227 .223(from the collection of different fragments of)J

916: 41 526 :M

917: 1.504 .15(different sizes )J

918: 112 526 :M

919: 3.012 .301(those that make up the most)J

920: 41 540 :M

921: -.114(appropriate analysis of an utterance.)A

922: 76 554 :M

923: .035 .004(DOP models have been shown to achieve)J

924: 41 568 :M

925: 6.543 .654(state-of-the-art parsing )J

926: 174 568 :M

927: 6.819 .682(performance on)J

928: 41 582 :M

929: 2.056 .206(benchmarks such )J

930: 128 582 :M

931: 2.615 .261(as the Wall Street Journal)J

932: 41 596 :M

933: -.016(corpus \(see Bod 2000a\). The original DOP model)A

934: 41 610 :M

935: 1.296 .13(in Bod \(1993\) was based on )J

936: 179 610 :M

937: -.018(utterance analyses)A

938: 41 624 :M

939: -.107(represented as surface trees, )A

940: 165 624 :M

941: .087 .009(and is equivalent to a)J

942: 41 638 :M

943: .482 .048(Stochastic Tree-Substitution )J

944: 172 638 :M

945: 2.08 .208(Grammar. But the)J

946: 41 652 :M

947: 2.226 .223(model has also been applied to several )J

948: 238 652 :M

949: -.15(other)A

950: 41 666 :M

951: 2.514 .251(grammatical )J

952: 105 666 :M

953: 2.89 .289(frameworks, e.g. Tree-Insertion)J

954: 41 680 :M

955: 3.487 .349(Grammar \(Hoogweg 2000\), Tree-Adjoining)J

956: 41 694 :M

957: 1.83 .183(Grammar )J

958: 90 694 :M

959: .751 .075(\(Neumann 1998\), Lexical-Functional)J

960: 41 708 :M

961: .089 .009(Grammar \(Bod & Kaplan )J

962: 158 708 :M

963: .436 .044(1998; Cormons 1999\),)J

964: 41 722 :M

965: 5.85 .585(Head-driven )J

966: 112 722 :M

967: 6.682 .668(Phrase Structure Grammar)J

968: 276 179 :M

969: 1.483 .148(\(Neumann & Flickinger )J

970: 393 179 :M

971: 1.476 .148(1999\), and Montague)J

972: 276 193 :M

973: 2.334 .233(Grammar \(Bonnema et al. 1997; Bod 1999\).)J

974: 276 207 :M

975: -.055(Most probability models for )A

976: 402 207 :M

977: .138 .014(DOP use the relative)J

978: 276 221 :M

979: 5.197 .52(frequency estimator to estimate )J

980: 455 221 :M

981: -.1(fragment)A

982: 276 235 :M

983: 4.068 .407(probabilities, although Bod )J

984: 423 235 :M

985: 3.477 .348(\(2000b\) trains)J

986: 276 249 :M

987: .012 .001(fragment probabilities by )J

988: 391 249 :M

989: .616 .062(a maximum likelihood)J

990: 276 263 :M

991: -.093(reestimation )A

992: 333 263 :M

993: .685 .068(procedure belonging to the class of)J

994: 276 277 :M

995: .573 .057(expectation-maximization algorithms. The )J

996: 471 277 :M

997: (DOP)S

998: 276 291 :M

999: -.086(model has also been )A

1000: 367 291 :M

1001: .434 .043(tested as a model for human)J

1002: 276 305 :M

1003: -.066(sentence processing \(Bod 2000d\).)A

1004: 311 319 :M

1005: 2.079 .208(This paper presents ongoing )J

1006: 453 319 :M

1007: 2.905 .291(work on)J

1008: 276 333 :M

1009: 1.88 .188(DOP models for Lexical-Functional )J

1010: 452 333 :M

1011: .143(Grammar)A

1012: 276 347 :M

1013: 1.008 .101(representations, known )J

1014: 386 347 :M

1015: 2.957 .296(as LFG-DOP \(Bod &)J

1016: 276 361 :M

1017: -.085(Kaplan )A

1018: 311 361 :M

1019: 1.087 .109(1998\). We develop a parser which uses)J

1020: 276 375 :M

1021: -.049(fragments from LFG-annotated sentences to )A

1022: 472 375 :M

1023: -.301(parse)A

1024: 276 389 :M

1025: .14 .014(new sentences, and we derive some experimental)J

1026: 276 403 :M

1027: 1.161 .116(properties of LFG-DOP on )J

1028: 406 403 :M

1029: .599 .06(two LFG-annotated)J

1030: 276 417 :M

1031: .195 .02(corpora: the Verbmobil )J

1032: 384 417 :M

1033: .684 .068(and Homecentre corpus.)J

1034: 276 431 :M

1035: .284 .028(The experiments show that the )J

1036: 417 431 :M

1037: 1.021 .102(DOP hypothesis,)J

1038: 276 445 :M

1039: 1.572 .157(which states that there is an )J

1040: 415 445 :M

1041: 1.31 .131(increase in parse)J

1042: 276 459 :M

1043: -.104(accuracy if larger fragments are taken into account)A

1044: 276 473 :M

1045: 1.728 .173(\(Bod 1998\), is confirmed )J

1046: 402 473 :M

1047: 2.421 .242(for LFG-DOP. We)J

1048: 276 487 :M

1049: 2.677 .268(report on an improved search technique for)J

1050: 276 501 :M

1051: 1.183 .118(estimating the most probable analysis. While a)J

1052: 276 515 :M

1053: 1.402 .14(Monte Carlo search converges provably to )J

1054: 481 515 :M

1055: -.719(the)A

1056: 276 529 :M

1057: 2.174 .217(most probable parse, a Viterbi )J

1058: 430 529 :M

1059: f4_11 sf

1060: (n)S

1061: 436 529 :M

1062: f3_11 sf

1063: 1.838 .184( best search)J

1064: 276 543 :M

1065: 3.12 .312(performs as well as Monte Carlo )J

1066: 452 543 :M

1067: 2.458 .246(while its)J

1068: 276 557 :M

1069: -.042(processing time is two orders of magnitude )A

1070: 468 557 :M

1071: -.195(faster.)A

1072: 276 571 :M

1073: .443 .044(We also show that LFG-DOP outperforms Tree)J

1074: 491 571 :M

1075: (-)S

1076: 276 585 :M

1077: -.072(DOP if evaluated on tree structures only.)A

1078: 276 609 :M

1079: f0_12 sf

1080: .561 .056(2  Summary of LFG-DOP)J

1081: 276 623 :M

1082: f3_11 sf

1083: .1 .01(In accordance with Bod \(1998\), a particular DOP)J

1084: 276 637 :M

1085: -.061(model is described by)A

1086: 276 661 :M

1087: f1_11 sf

1088: <B7>S

1089: 282 661 :M

1090: f3_11 sf

1091: -.018(\312a definition of a well-formed )A

1092: 415 661 :M

1093: f4_11 sf

1094: -.043(representation for)A

1095: 276 676 :M

1096: -.106(utterance analyses,)A

1097: 276 691 :M

1098: f1_11 sf

1099: <B7>S

1100: 282 691 :M

1101: f3_11 sf

1102: .048 .005(\312a set of )J

1103: f4_11 sf

1104: .18 .018(decomposition operations)J

1105: 435 691 :M

1106: f3_11 sf

1107: ( that )S

1108: 458 691 :M

1109: -.018(divide a)A

1110: 276 706 :M

1111: -.126(given utterance analysis into a set of )A

1112: 435 706 :M

1113: f4_11 sf

1114: -.078(fragments)A

1115: f3_11 sf

1116: (,)S

1117: endp

1118: %%Page: 2 2

1119: %%BeginPageSetup

1120: initializepage

1121: (rens; page: 2 of 8)setjob

1122: %%EndPageSetup

1123: -29 -30 :T

1124: gS 29 30 538 781 rC

1125: 70 97 :M

1126: f1_11 sf

1127: <B7>S

1128: 76 97 :M

1129: f3_11 sf

1130: .077 .008(\312a set of )J

1131: f4_11 sf

1132: .277 .028(composition operations)J

1133: 219 97 :M

1134: f3_11 sf

1135: .479 .048( by which )J

1136: 268 97 :M

1137: -.053(such)A

1138: 70 112 :M

1139: 2.803 .28(fragments may be recombined )J

1140: 226 112 :M

1141: 3.252 .325(to derive an)J

1142: 70 126 :M

1143: -.112(analysis of a new utterance, and)A

1144: 70 141 :M

1145: f1_11 sf

1146: <B7>S

1147: 76 141 :M

1148: f3_11 sf

1149: -.056(\312a definition of a )A

1150: f4_11 sf

1151: -.065(probability model)A

1152: f3_11 sf

1153: -.062( that indicates)A

1154: 70 156 :M

1155: -.002(how the probability of a new utterance analysis is)A

1156: 70 170 :M

1157: -.141(computed.)A

1158: 70 193 :M

1159: 7.422 .742(In defining a DOP )J

1160: 198 193 :M

1161: 7.315 .732(model for LFG)J

1162: 70 207 :M

1163: .871 .087(representations, Bod & )J

1164: 180 207 :M

1165: 1.233 .123(Kaplan \(1998\) give the)J

1166: 70 221 :M

1167: -.012(following settings for DOP's four parameters.)A

1168: 70 244 :M

1169: f0_11 sf

1170: (2.1  Representations)S

1171: 70 258 :M

1172: f3_11 sf

1173: 3.159 .316(The representations used by )J

1174: 218 258 :M

1175: 3.196 .32(LFG-DOP are)J

1176: 70 272 :M

1177: 1.589 .159(directly taken from LFG: they consist of )J

1178: 269 272 :M

1179: 2.071 .207(a c)J

1180: 285 272 :M

1181: (-)S

1182: 70 287 :M

1183: -.055(structure, an f-structure and a mapping )A

1184: 242 287 :M

1185: f1_11 sf

1186: .111(f)A

1187: f3_11 sf

1188: .449 .045( between)J

1189: 70 302 :M

1190: 1.003 .1(them. )J

1191: 100 302 :M

1192: 1.509 .151(The following figure shows an example)J

1193: 70 316 :M

1194: -.034(representation for )A

1195: 151 316 :M

1196: f4_11 sf

1197: .485 .049(Kim eats)J

1198: f3_11 sf

1199: .292 .029(. \(We leave )J

1200: 246 316 :M

1201: 1.163 .116(out some)J

1202: 70 330 :M

1203: -.092(features to keep the example simple.\))A

1204: 112 367 6 9 rC

1205: gS

1206: .599 .597 scale

1207: 191.533 621.866 :T

1208: -191.533 -621.866 :T

1209: 188.533 626.866 :M

1210: f3_12 sf

1211: (S)S

1212: gR

1213: gR

1214: gS 87 342 184 71 rC

1215: gR

1216: gS 29 30 538 781 rC

1217: 102 384 :M

1218: 109.332 377.994 113 375 113 375 :C

1219: 113 375 114.998 376.328 119 379 :C

1220: 122.998 381.661 125 383 125 383 :C

1221: stroke

1222: 98 384 11 9 rC

1223: gS

1224: .599 .597 scale

1225: 172.175 650.36 :T

1226: -172.175 -650.36 :T

1227: 165.175 655.36 :M

1228: f3_12 sf

1229: (NP)S

1230: gR

1231: gR

1232: gS 121 384 10 9 rC

1233: gS

1234: .599 .597 scale

1235: 208.88 650.36 :T

1236: -208.88 -650.36 :T

1237: 201.88 655.36 :M

1238: f3_12 sf

1239: (VP)S

1240: gR

1241: gR

1242: gS 97 401 13 9 rC

1243: gS

1244: .599 .597 scale

1245: 171.838 678.854 :T

1246: -171.838 -678.854 :T

1247: 161.838 683.854 :M

1248: f3_12 sf

1249: (Kim)S

1250: gR

1251: gR

1252: gS 121 400 12 9 rC

1253: gS

1254: .599 .597 scale

1255: 210.88 678.854 :T

1256: -210.88 -678.854 :T

1257: 201.88 683.854 :M

1258: f3_12 sf

1259: (eats)S

1260: gR

1261: gR

1262: gS 87 342 184 71 rC

1263: -1 -1 126 401 1 1 125 393 @b

1264: 226 359 40 9 rC

1265: gS

1266: .599 .597 scale

1267: 409.734 608.457 :T

1268: -409.734 -608.457 :T

1269: 378.734 613.457 :M

1270: f3_10 sf

1271: (PRED)S

1272: gR

1273: gS

1274: .599 .597 scale

1275: 409.734 608.457 :T

1276: -409.734 -608.457 :T

1277: 403.76 613.457 :M

1278: f3_12 sf

1279: (   'Kim')S

1280: gR

1281: gR

1282: gS 226 370 36 9 rC

1283: gS

1284: .599 .597 scale

1285: 406.734 626.894 :T

1286: -406.734 -626.894 :T

1287: 378.734 631.894 :M

1288: f3_10 sf

1289: (NUM\011    SG)S

1290: gR

1291: gS

1292: .599 .597 scale

1293: 406.734 626.894 :T

1294: -406.734 -626.894 :T

1295: 427.119 631.894 :M

1296: f3_12 sf

1297: ( )S

1298: gR

1299: gR

1300: gS 87 342 184 71 rC

1301: gR

1302: gS 29 30 538 781 rC

1303: 223 358 :M

1304: 222.33 358 222 358 222 358 :C

1305: 222 358 222 361.994 222 370 :C

1306: 222 377.994 222 382 222 382 :C

1307: 222 382 222.163 382 222.5 382 :C

1308: 222.83 382 223 382 223 382 :C

1309: stroke

1310: 194 365 18 7 rC

1311: gS

1312: .599 .597 scale

1313: 336.676 617.838 :T

1314: -336.676 -617.838 :T

1315: 323.676 621.838 :M

1316: f3_10 sf

1317: (SUBJ)S

1318: gR

1319: gR

1320: gS 193 387 50 8 rC

1321: gS

1322: .599 .597 scale

1323: 364.676 654.712 :T

1324: -364.676 -654.712 :T

1325: 323.676 658.712 :M

1326: f3_10 sf

1327: (TENSE      PRES)S

1328: gR

1329: gR

1330: gS 194 400 62 9 rC

1331: gS

1332: .599 .597 scale

1333: 373.676 677.178 :T

1334: -373.676 -677.178 :T

1335: 323.676 682.178 :M

1336: f3_10 sf

1337: (PRED)S

1338: gR

1339: gS

1340: .599 .597 scale

1341: 373.676 677.178 :T

1342: -373.676 -677.178 :T

1343: 350.371 682.178 :M

1344: f3_12 sf

1345: (      'eat\()S

1346: gR

1347: gS

1348: .599 .597 scale

1349: 373.676 677.178 :T

1350: -373.676 -677.178 :T

1351: 388.745 682.178 :M

1352: f3_10 sf

1353: (SUBJ)S

1354: f3_12 sf

1355: (\)')S

1356: gR

1357: gR

1358: gS 87 342 184 71 rC

1359: gR

1360: gS 29 30 538 781 rC

1361: 190 356 :M

1362: 188.664 356 188 356 188 356 :C

1363: 188 356 188 365.327 188 384 :C

1364: 188 402.661 188 412 188 412 :C

1365: 188 412 188.33 412 189 412 :C

1366: 189.664 412 190 412 190 412 :C

1367: stroke

1368: 87 342 184 71 rC

1369: gR

1370: gS 29 30 538 781 rC

1371: 263 358 :M

1372: 263.663 358 264 358 264 358 :C

1373: 264 358 264 361.994 264 370 :C

1374: 264 377.994 264 382 264 382 :C

1375: 264 382 263.829 382 263.5 382 :C

1376: 263.163 382 263 382 263 382 :C

1377: stroke

1378: 87 342 184 71 rC

1379: gR

1380: gS 29 30 538 781 rC

1381: 269 355 :M

1382: 269.663 355 270 355 270 355 :C

1383: 270 355 270 364.327 270 383 :C

1384: 270 401.661 270 411 270 411 :C

1385: 270 411 269.829 411 269.5 411 :C

1386: 269.163 411 269 411 269 411 :C

1387: stroke

1388: 87 342 184 71 rC

1389: -1 -1 104 401 1 1 103 392 @b

1390: 180 270 99 33 181 403 @n

1391: gR

1392: gS 29 30 538 781 rC

1393: np 186 386 :M

1394: 179 388 :L

1395: 179 386 :L

1396: 179 384 :L

1397: 186 386 :L

1398: eofill

1399: 182.5 387 :M

1400: 180.164 387.661 179 388 179 388 :C

1401: 179 388 179 387.661 179 387 :C

1402: 179 386.327 179 386 179 386 :C

1403: 179 386 179 385.661 179 385 :C

1404: 179 384.327 179 384 179 384 :C

1405: 179 384 180.164 384.327 182.5 385 :C

1406: 184.831 385.661 186 386 186 386 :C

1407: 186 386 184.831 386.327 182.5 387 :C

1408: :K

1409: eofill

1410: 87 342 184 71 rC

1411: 179 387 -1 1 186 386 1 179 386 @a

1412: 180 270 193 83 194 385 @n

1413: -90 0 50 15 192.5 351 @n

1414: gR

1415: gS 29 30 538 781 rC

1416: np 221 355 :M

1417: 215 350 :L

1418: 216 349 :L

1419: 218 347 :L

1420: 221 355 :L

1421: eofill

1422: 218 352.5 :M

1423: 215.997 350.828 215 350 215 350 :C

1424: 215 350 215.163 349.828 215.5 349.5 :C

1425: 215.83 349.161 216 349 216 349 :C

1426: 216 349 216.33 348.661 217 348 :C

1427: 217.663 347.328 218 347 218 347 :C

1428: 218 347 218.497 348.328 219.5 351 :C

1429: 220.497 353.661 221 355 221 355 :C

1430: 221 355 219.997 354.161 218 352.5 :C

1431: :K

1432: eofill

1433: 87 342 184 71 rC

1434: 216 350 -1 1 219 350 1 216 349 @a

1435: 90 180 13 28 95 388.5 @n

1436: 180 270 206 90 191.5 388.5 @n

1437: -1 -1 134 388 1 1 179 386 @b

1438: 90 180 119 21 179 375 @n

1439: gR

1440: gS 29 30 538 781 rC

1441: 161 432 :M

1442: f3_10 sf

1443: .822 .082(Figure 1)J

1444: 70 455 :M

1445: f3_11 sf

1446: 3.043 .304(Bod & )J

1447: 111 455 :M

1448: 2.562 .256(Kaplan also introduce the notion of)J

1449: 70 469 :M

1450: .076 .008(accessibility which they later use for defining )J

1451: 275 469 :M

1452: -.719(the)A

1453: 70 483 :M

1454: -.014(decomposition operations of LFG-DOP:)A

1455: 78 507 :M

1456: -.084(An f-structure unit )A

1457: 162 507 :M

1458: f4_11 sf

1459: -.364(f)A

1460: f3_11 sf

1461: -.509( is )A

1462: 177 507 :M

1463: f1_11 sf

1464: .371(f)A

1465: f3_11 sf

1466: (-)S

1467: 188 507 :M

1468: f4_11 sf

1469: -.097(accessible)A

1470: f3_11 sf

1471: -.106( from a )A

1472: 267 507 :M

1473: -.127(node)A

1474: 78 523 :M

1475: f4_11 sf

1476: (n)S

1477: 84 523 :M

1478: f3_11 sf

1479: -.151( iff either )A

1480: 127 523 :M

1481: f4_11 sf

1482: .2(n)A

1483: f3_11 sf

1484: .203 .02( is )J

1485: 147 523 :M

1486: f1_11 sf

1487: .119(f)A

1488: f3_11 sf

1489: .313 .031(-linked to )J

1490: 200 523 :M

1491: f4_11 sf

1492: (f)S

1493: 203 523 :M

1494: f3_11 sf

1495: .378 .038( \(that is, )J

1496: 244 523 :M

1497: f4_11 sf

1498: .406 .041(f )J

1499: f3_11 sf

1500: .684 .068(= )J

1501: 261 523 :M

1502: f1_11 sf

1503: .371(f)A

1504: f3_11 sf

1505: <28>S

1506: 272 523 :M

1507: f4_11 sf

1508: .152(n)A

1509: f3_11 sf

1510: .231 .023(\) \))J

1511: 78 539 :M

1512: .075 .007(or )J

1513: 91 539 :M

1514: f4_11 sf

1515: (f)S

1516: 94 539 :M

1517: f3_11 sf

1518: .133 .013( is contained within )J

1519: 185 539 :M

1520: f1_11 sf

1521: .371(f)A

1522: f3_11 sf

1523: <28>S

1524: 196 539 :M

1525: f4_11 sf

1526: .156(n)A

1527: f3_11 sf

1528: .291 .029(\) \(that is, )J

1529: 246 539 :M

1530: .794 .079(there is a)J

1531: 78 555 :M

1532: -.111(chain of attributes that leads from )A

1533: 226 555 :M

1534: f1_11 sf

1535: .371(f)A

1536: f3_11 sf

1537: <28>S

1538: 237 555 :M

1539: f4_11 sf

1540: -.288(n)A

1541: f3_11 sf

1542: -.232(\) to )A

1543: 260 555 :M

1544: f4_11 sf

1545: (f)S

1546: 263 555 :M

1547: f3_11 sf

1548: -.41(\).)A

1549: 70 579 :M

1550: .724 .072(According to the )J

1551: 151 579 :M

1552: 1.043 .104(LFG representation theory, c)J

1553: 285 579 :M

1554: (-)S

1555: 70 593 :M

1556: 1.136 .114(structures and )J

1557: 139 593 :M

1558: 1.557 .156(f-structures must satisfy certain)J

1559: 70 607 :M

1560: 5.114 .511(formal well-formedness conditions. A )J

1561: 280 607 :M

1562: -1.543(c-)A

1563: 70 621 :M

1564: -.104(structure/f-structure pair is )A

1565: 188 621 :M

1566: (a )S

1567: f4_11 sf

1568: (valid)S

1569: f3_11 sf

1570: -.011( LFG represent)A

1571: 285 621 :M

1572: (-)S

1573: 70 635 :M

1574: 3.6 .36(ation only if )J

1575: 143 635 :M

1576: 3.077 .308(it satisfies the Nonbranching)J

1577: 70 649 :M

1578: 1.001 .1(Dominance, Uniqueness, Coherence )J

1579: 241 649 :M

1580: 1.949 .195(and Com-)J

1581: 70 663 :M

1582: -.053(pleteness conditions \(Kaplan & Bresnan 1982\).)A

1583: 70 686 :M

1584: f0_11 sf

1585: .102 .01(2.2  Decomposition operations and Fragments)J

1586: 70 700 :M

1587: f3_11 sf

1588: 4.729 .473(The )J

1589: 97 700 :M

1590: 5.079 .508(fragments for LFG-DOP consist of)J

1591: 70 715 :M

1592: 3.267 .327(connected subtrees whose nodes are )J

1593: 260 715 :M

1594: 4.745 .474(in )J

1595: 278 715 :M

1596: f1_11 sf

1597: .371(f)A

1598: f3_11 sf

1599: (-)S

1600: 70 730 :M

1601: .518 .052(correspondence with the )J

1602: 184 730 :M

1603: .412 .041(correponding sub-units)J

1604: 70 744 :M

1605: 1.525 .152(of f-structures. To give a precise )J

1606: 231 744 :M

1607: .601 .06(definition of)J

1608: 305 96 :M

1609: -.053(LFG-DOP fragments, it is convenient to recall the)A

1610: 305 110 :M

1611: 2.384 .238(decomposition )J

1612: 379 110 :M

1613: 4.022 .402(operations employed by the)J

1614: 305 124 :M

1615: .803 .08(orginal DOP model which is also known as the)J

1616: 305 138 :M

1617: -.022("Tree-DOP" model \(Bod 1993, 1998\):)A

1618: 313 161 :M

1619: -.065<283129CACA>A

1620: f4_11 sf

1621: -.093(Root)A

1622: f3_11 sf

1623: -.079(: the )A

1624: 374 161 :M

1625: f4_11 sf

1626: -.12(Root )A

1627: f3_11 sf

1628: -.117(operation selects any node )A

1629: 514 161 :M

1630: -.16(of)A

1631: 313 175 :M

1632: .982 .098(a tree to )J

1633: 356 175 :M

1634: 1.713 .171(be the root of the new subtree and)J

1635: 313 189 :M

1636: -.1(erases all nodes except the selected node and the)A

1637: 313 203 :M

1638: -.069(nodes it dominates.)A

1639: 313 226 :M

1640: -.039<283229CACA>A

1641: f4_11 sf

1642: -.057(Frontier)A

1643: 369 226 :M

1644: f3_11 sf

1645: 4.722 .472(: )J

1646: 382 226 :M

1647: 4.47 .447(the )J

1648: 405 226 :M

1649: f4_11 sf

1650: 3.878 .388(Frontier )J

1651: 453 226 :M

1652: f3_11 sf

1653: 2.076 .208(operation then)J

1654: 313 240 :M

1655: .507 .051(chooses a set \(possibly empty\) of nodes )J

1656: 497 240 :M

1657: .17 .017(in the)J

1658: 313 254 :M

1659: -.065(new subtree different from its root and erases all)A

1660: 313 268 :M

1661: -.042(subtrees dominated by the chosen nodes.)A

1662: 305 291 :M

1663: 2.377 .238(Bod & Kaplan extend )J

1664: 419 291 :M

1665: 2.095 .21(Tree-DOP's )J

1666: 479 291 :M

1667: f4_11 sf

1668: .473(Root)A

1669: f3_11 sf

1670: 1.306 .131( and)J

1671: 305 305 :M

1672: f4_11 sf

1673: -.039(Frontier)A

1674: 343 305 :M

1675: f3_11 sf

1676: .151 .015( operations so that they also apply to the)J

1677: 305 319 :M

1678: .149 .015(nodes of the c-structure in LFG, while respecting)J

1679: 305 333 :M

1680: -.097(the principles of c/f-structure correspondence.)A

1681: 340 347 :M

1682: 2.059 .206(When a )J

1683: 383 347 :M

1684: 2.284 .228(node is selected by the )J

1685: 503 347 :M

1686: f4_11 sf

1687: .075(Root)A

1688: 305 361 :M

1689: f3_11 sf

1690: -.204(operation, )A

1691: 351 361 :M

1692: .204 .02(all nodes outside of that node's subtree)J

1693: 305 375 :M

1694: 1.404 .14(are )J

1695: 324 375 :M

1696: 2.256 .226(erased, just as in Tree-DOP. Further, for)J

1697: 305 390 :M

1698: .726 .073(LFG-DOP, all )J

1699: 374 390 :M

1700: f1_11 sf

1701: 2.29 .229(f )J

1702: 386 390 :M

1703: f3_11 sf

1704: .587 .059(links leaving the erased nodes)J

1705: 305 405 :M

1706: .206 .021(are removed and all f-structure )J

1707: 446 405 :M

1708: .72 .072(units that are not)J

1709: 305 420 :M

1710: f1_11 sf

1711: -.06(f)A

1712: f3_11 sf

1713: -.048(-accessible from the remaining nodes are )A

1714: 493 420 :M

1715: -.139(erased.)A

1716: 305 435 :M

1717: .388 .039(For example, if )J

1718: 378 435 :M

1719: f4_11 sf

1720: .038(Root)A

1721: f3_11 sf

1722: .091 .009( selects )J

1723: 435 435 :M

1724: 1.11 .111(the NP in figure 1,)J

1725: 305 449 :M

1726: -.038(then the f-structure corresponding to the S node is)A

1727: 305 463 :M

1728: -.061(erased, giving figure 2 as a possible fragment:)A

1729: 350 504 11 9 rC

1730: gS

1731: .63 .596 scale

1732: 563.923 852.744 :T

1733: -563.923 -852.744 :T

1734: 556.923 857.744 :M

1735: f3_12 sf

1736: (NP)S

1737: gR

1738: gR

1739: gS 348 521 15 9 rC

1740: gS

1741: .63 .596 scale

1742: 563.749 882.958 :T

1743: -563.749 -882.958 :T

1744: 553.749 887.958 :M

1745: f3_12 sf

1746: (Kim)S

1747: gR

1748: gR

1749: gS 446 492 42 9 rC

1750: gS

1751: .63 .596 scale

1752: 741.243 832.601 :T

1753: -741.243 -832.601 :T

1754: 709.243 837.601 :M

1755: f3_10 sf

1756: (PRED)S

1757: gR

1758: gS

1759: .63 .596 scale

1760: 741.243 832.601 :T

1761: -741.243 -832.601 :T

1762: 734.63 837.601 :M

1763: f3_12 sf

1764: (   'Kim')S

1765: gR

1766: gR

1767: gS 446 503 38 9 rC

1768: gS

1769: .63 .596 scale

1770: 737.243 851.065 :T

1771: -737.243 -851.065 :T

1772: 709.243 856.065 :M

1773: f3_10 sf

1774: (NUM\011    SG)S

1775: gR

1776: gS

1777: .63 .596 scale

1778: 737.243 851.065 :T

1779: -737.243 -851.065 :T

1780: 758.43 856.065 :M

1781: f3_12 sf

1782: ( )S

1783: gR

1784: gR

1785: gS 339 475 150 56 rC

1786: gR

1787: gS 29 30 538 781 rC

1788: 444 491 :M

1789: 442.66 491 442 491 442 491 :C

1790: 442 491 442 494.992 442 503 :C

1791: 442 510.992 442 515 442 515 :C

1792: 442 515 442.327 515 443 515 :C

1793: 443.66 515 444 515 444 515 :C

1794: stroke

1795: 339 475 150 56 rC

1796: gR

1797: gS 29 30 538 781 rC

1798: 486 491 :M

1799: 486.659 491 487 491 487 491 :C

1800: 487 491 487 494.992 487 503 :C

1801: 487 510.992 487 515 487 515 :C

1802: 487 515 486.826 515 486.5 515 :C

1803: 486.159 515 486 515 486 515 :C

1804: stroke

1805: 339 475 150 56 rC

1806: -1 -1 356 521 1 1 355 513 @b

1807: 180 270 129 57 413 505 @n

1808: -90 0 53 15 412 484 @n

1809: np 442 488 :M

1810: 435 483 :L

1811: 437 482 :L

1812: 439 480 :L

1813: 442 488 :L

1814: eofill

1815: 437 483 -1 1 439 483 1 437 482 @a

1816: 90 180 14 28 347.5 512.5 @n

1817: 180 270 140 74 410.5 513.5 @n

1818: gR

1819: gS 29 30 538 781 rC

1820: 396 550 :M

1821: f3_10 sf

1822: .822 .082(Figure 2)J

1823: 305 573 :M

1824: f3_11 sf

1825: .689 .069(In addition the )J

1826: 376 573 :M

1827: f4_11 sf

1828: .326(Root)A

1829: f3_11 sf

1830: .172 .017( )J

1831: 402 573 :M

1832: 1.065 .107(operation deletes from the)J

1833: 305 587 :M

1834: .412 .041(remaining f-structure all semantic forms that are)J

1835: 305 601 :M

1836: -.225(local )A

1837: 329 601 :M

1838: .783 .078(to f-structures that correspond to erased c)J

1839: 520 601 :M

1840: (-)S

1841: 305 615 :M

1842: .207 .021(structure nodes, and it thereby also maintains )J

1843: 510 615 :M

1844: -.719(the)A

1845: 305 629 :M

1846: (fundamental two-way connection )S

1847: 456 629 :M

1848: .33 .033(between words)J

1849: 305 643 :M

1850: .121 .012(and meanings. Thus, if )J

1851: 410 643 :M

1852: f4_11 sf

1853: .096(Root)A

1854: f3_11 sf

1855: .279 .028( selects the VP node)J

1856: 305 657 :M

1857: 1.749 .175(so that the NP is erased, the subject )J

1858: 485 657 :M

1859: -.156(semantic)A

1860: 305 671 :M

1861: -.03(form "Kim" is also deleted:)A

1862: 336 704 11 9 rC

1863: gS

1864: .642 .6 scale

1865: 530.154 1180.078 :T

1866: -530.154 -1180.078 :T

1867: 523.154 1185.078 :M

1868: f3_12 sf

1869: (VP)S

1870: gR

1871: gR

1872: gS 336 721 13 9 rC

1873: gS

1874: .642 .6 scale

1875: 532.154 1208.414 :T

1876: -532.154 -1208.414 :T

1877: 523.154 1213.414 :M

1878: f3_12 sf

1879: (eats)S

1880: gR

1881: gR

1882: gS 335 683 158 51 rC

1883: -1 -1 342 722 1 1 341 713 @b

1884: 449 690 39 9 rC

1885: gS

1886: .642 .6 scale

1887: 728.096 1156.743 :T

1888: -728.096 -1156.743 :T

1889: 699.096 1161.743 :M

1890: f3_10 sf

1891: (NUM\011    SG)S

1892: gR

1893: gS

1894: .642 .6 scale

1895: 728.096 1156.743 :T

1896: -728.096 -1156.743 :T

1897: 748.921 1161.743 :M

1898: f3_12 sf

1899: ( )S

1900: gR

1901: gR

1902: gS 335 683 158 51 rC

1903: gR

1904: gS 29 30 538 781 rC

1905: 445 689 :M

1906: 444.327 689 444 689 444 689 :C

1907: 444 689 444 691.156 444 695.5 :C

1908: 444 699.823 444 702 444 702 :C

1909: 444 702 444.16 702 444.5 702 :C

1910: 444.827 702 445 702 445 702 :C

1911: stroke

1912: 413 691 20 8 rC

1913: gS

1914: .642 .6 scale

1915: 658.601 1157.743 :T

1916: -658.601 -1157.743 :T

1917: 644.601 1161.743 :M

1918: f3_10 sf

1919: (SUBJ)S

1920: gR

1921: gR

1922: gS 413 708 55 7 rC

1923: gS

1924: .642 .6 scale

1925: 685.601 1186.079 :T

1926: -685.601 -1186.079 :T

1927: 644.601 1190.079 :M

1928: f3_10 sf

1929: (TENSE      PRES)S

1930: gR

1931: gR

1932: gS 414 720 67 9 rC

1933: gS

1934: .642 .6 scale

1935: 697.158 1206.747 :T

1936: -697.158 -1206.747 :T

1937: 646.158 1211.747 :M

1938: f3_10 sf

1939: (PRED)S

1940: f3_12 sf

1941: (      'eat\()S

1942: gR

1943: gS

1944: .642 .6 scale

1945: 697.158 1206.747 :T

1946: -697.158 -1206.747 :T

1947: 709.995 1211.747 :M

1948: f3_10 sf

1949: (SUBJ)S

1950: gR

1951: gS

1952: .642 .6 scale

1953: 697.158 1206.747 :T

1954: -697.158 -1206.747 :T

1955: 733.351 1211.747 :M

1956: f3_12 sf

1957: (\)')S

1958: gR

1959: gR

1960: gS 335 683 158 51 rC

1961: gR

1962: gS 29 30 538 781 rC

1963: 411 685 :M

1964: 409.66 685 409 685 409 685 :C

1965: 409 685 409 692.989 409 709 :C

1966: 409 724.989 409 733 409 733 :C

1967: 409 733 409.327 733 410 733 :C

1968: 410.66 733 411 733 411 733 :C

1969: stroke

1970: 335 683 158 51 rC

1971: gR

1972: gS 29 30 538 781 rC

1973: 485 688 :M

1974: 486.326 688 487 688 487 688 :C

1975: 487 688 487 690.156 487 694.5 :C

1976: 487 698.823 487 701 487 701 :C

1977: 487 701 486.659 701 486 701 :C

1978: 485.326 701 485 701 485 701 :C

1979: stroke

1980: 335 683 158 51 rC

1981: 180 270 106 34 400.5 723.5 @n

1982: gR

1983: gS 29 30 538 781 rC

1984: np 406 706 :M

1985: 398 708 :L

1986: 398 706 :L

1987: 398 705 :L

1988: 406 706 :L

1989: eofill

1990: 402 707 :M

1991: 399.327 707.656 398 708 398 708 :C

1992: 398 708 398 707.656 398 707 :C

1993: 398 706.323 398 706 398 706 :C

1994: 398 706 398 705.823 398 705.5 :C

1995: 398 705.156 398 705 398 705 :C

1996: 398 705 399.327 705.156 402 705.5 :C

1997: 404.661 705.823 406 706 406 706 :C

1998: 406 706 404.661 706.323 402 707 :C

1999: :K

2000: eofill

2001: 335 683 158 51 rC

2002: 398 707 -1 1 406 706 1 398 706 @a

2003: -1 -1 349 709 1 1 399 706 @b

2004: gR

2005: gS 29 30 538 781 rC

2006: 490 684 :M

2007: 491.326 684 492 684 492 684 :C

2008: 492 684 492 692.156 492 708.5 :C

2009: 492 724.823 492 733 492 733 :C

2010: 492 733 491.659 733 491 733 :C

2011: 490.326 733 490 733 490 733 :C

2012: stroke

2013: 396 753 :M

2014: f3_10 sf

2015: .822 .082(Figure 3)J

2016: endp

2017: %%Page: 3 3

2018: %%BeginPageSetup

2019: initializepage

2020: (rens; page: 3 of 8)setjob

2021: %%EndPageSetup

2022: -29 -30 :T

2023: gS 29 30 538 781 rC

2024: 70 96 :M

2025: f3_11 sf

2026: 1.022 .102(As with Tree-DOP, the )J

2027: 183 96 :M

2028: f4_11 sf

2029: .094(Frontier)A

2030: f3_11 sf

2031: .315 .031( operation )J

2032: 270 96 :M

2033: -.313(then)A

2034: 70 110 :M

2035: 1.473 .147(selects a set )J

2036: 132 110 :M

2037: 1.956 .196(of frontier nodes and deletes all)J

2038: 70 124 :M

2039: 3.036 .304(subtrees they dominate. )J

2040: 194 124 :M

2041: 3.637 .364(Like )J

2042: 223 124 :M

2043: f4_11 sf

2044: 1.194(Root)A

2045: f3_11 sf

2046: 2.633 .263(, it also)J

2047: 70 139 :M

2048: 1.549 .155(removes the )J

2049: 132 139 :M

2050: f1_11 sf

2051: .758(f)A

2052: f3_11 sf

2053: 1.793 .179( links of the deleted nodes )J

2054: 273 139 :M

2055: -.441(and)A

2056: 70 154 :M

2057: .125 .012(erases any semantic form that corresponds to any)J

2058: 70 168 :M

2059: -.065(of those nodes. For instance, if )A

2060: 207 168 :M

2061: .384 .038(the NP in figure 1)J

2062: 70 182 :M

2063: .285 .029(is selected as a frontier node, )J

2064: 204 182 :M

2065: f4_11 sf

2066: .025(Frontier)A

2067: f3_11 sf

2068: .063 .006( erases )J

2069: 275 182 :M

2070: -.719(the)A

2071: 70 196 :M

2072: -.042(predicate "Kim" from the fragment:)A

2073: 116 267 13 9 rC

2074: gS

2075: .611 .597 scale

2076: 198.756 454.255 :T

2077: -198.756 -454.255 :T

2078: 189.756 459.255 :M

2079: f3_12 sf

2080: (eats)S

2081: gR

2082: gR

2083: gS 109 233 5 9 rC

2084: gS

2085: .611 .597 scale

2086: 181.305 397.267 :T

2087: -181.305 -397.267 :T

2088: 178.305 402.267 :M

2089: f3_12 sf

2090: (S)S

2091: gR

2092: gR

2093: gS 92 208 173 71 rC

2094: gR

2095: gS 29 30 538 781 rC

2096: 98 250 :M

2097: 105.998 243.996 110 241 110 241 :C

2098: 110 241 111.832 242.33 115.5 245 :C

2099: 119.165 247.663 121 249 121 249 :C

2100: stroke

2101: 94 250 10 9 rC

2102: gS

2103: .611 .597 scale

2104: 160.767 425.761 :T

2105: -160.767 -425.761 :T

2106: 153.767 430.761 :M

2107: f3_12 sf

2108: (NP)S

2109: gR

2110: gR

2111: gS 116 250 11 9 rC

2112: gS

2113: .611 .597 scale

2114: 198.391 425.761 :T

2115: -198.391 -425.761 :T

2116: 191.391 430.761 :M

2117: f3_12 sf

2118: (VP)S

2119: gR

2120: gR

2121: gS 92 208 173 71 rC

2122: -1 -1 122 267 1 1 121 259 @b

2123: 225 232 35 8 rC

2124: gS

2125: .611 .597 scale

2126: 396.06 395.915 :T

2127: -396.06 -395.915 :T

2128: 368.06 398.915 :M

2129: f3_10 sf

2130: (NUM\011   SG)S

2131: f3_12 sf

2132: ( )S

2133: gR

2134: gR

2135: gS 192 231 17 8 rC

2136: gS

2137: .611 .597 scale

2138: 327.714 393.239 :T

2139: -327.714 -393.239 :T

2140: 315.714 397.239 :M

2141: f3_10 sf

2142: (SUBJ)S

2143: gR

2144: gR

2145: gS 92 208 173 71 rC

2146: gR

2147: gS 29 30 538 781 rC

2148: 188 225 :M

2149: 187.33 225 187 225 187 225 :C

2150: 187 225 187 233.829 187 251.5 :C

2151: 187 269.163 187 278 187 278 :C

2152: 187 278 187.164 278 187.5 278 :C

2153: 187.83 278 188 278 188 278 :C

2154: stroke

2155: 92 208 173 71 rC

2156: gR

2157: gS 29 30 538 781 rC

2158: 263 224 :M

2159: 263.663 224 264 224 264 224 :C

2160: 264 224 264 232.829 264 250.5 :C

2161: 264 268.163 264 277 264 277 :C

2162: 264 277 263.829 277 263.5 277 :C

2163: 263.163 277 263 277 263 277 :C

2164: stroke

2165: 92 208 173 71 rC

2166: 180 270 101 33 178 269 @n

2167: gR

2168: gS 29 30 538 781 rC

2169: np 183 252 :M

2170: 176 254 :L

2171: 176 252 :L

2172: 176 250 :L

2173: 183 252 :L

2174: eofill

2175: 179.5 253 :M

2176: 177.164 253.663 176 254 176 254 :C

2177: 176 254 176 253.663 176 253 :C

2178: 176 252.329 176 252 176 252 :C

2179: 176 252 176 251.663 176 251 :C

2180: 176 250.329 176 250 176 250 :C

2181: 176 250 177.164 250.329 179.5 251 :C

2182: 181.831 251.663 183 252 183 252 :C

2183: 183 252 181.831 252.329 179.5 253 :C

2184: :K

2185: eofill

2186: 92 208 173 71 rC

2187: 176 253 -1 1 183 252 1 176 252 @a

2188: 180 270 195 83 191 251 @n

2189: -90 0 59 22 190 220.5 @n

2190: gR

2191: gS 29 30 538 781 rC

2192: np 223 225 :M

2193: 217 220 :L

2194: 219 219 :L

2195: 220 218 :L

2196: 223 225 :L

2197: eofill

2198: 220 222.5 :M

2199: 217.997 220.83 217 220 217 220 :C

2200: 217 220 217.33 219.83 218 219.5 :C

2201: 218.663 219.163 219 219 219 219 :C

2202: 219 219 219.163 218.83 219.5 218.5 :C

2203: 219.83 218.163 220 218 220 218 :C

2204: 220 218 220.497 219.163 221.5 221.5 :C

2205: 222.497 223.83 223 225 223 225 :C

2206: 223 225 221.997 224.163 220 222.5 :C

2207: :K

2208: eofill

2209: 92 208 173 71 rC

2210: -1 -1 220 221 1 1 219 219 @b

2211: -1 -1 130 254 1 1 176 252 @b

2212: 90 180 121 21 176 241 @n

2213: gR

2214: gS 29 30 538 781 rC

2215: 223 230 :M

2216: 222.33 230 222 230 222 230 :C

2217: 222 230 222 231.663 222 235 :C

2218: 222 238.33 222 240 222 240 :C

2219: 222 240 222.163 240 222.5 240 :C

2220: 222.83 240 223 240 223 240 :C

2221: stroke

2222: 92 208 173 71 rC

2223: gR

2224: gS 29 30 538 781 rC

2225: 256 230 :M

2226: 257.329 230 258 230 258 230 :C

2227: 258 230 258 231.663 258 235 :C

2228: 258 238.33 258 240 258 240 :C

2229: 258 240 257.663 240 257 240 :C

2230: 256.329 240 256 240 256 240 :C

2231: stroke

2232: 192 249 52 7 rC

2233: gS

2234: .611 .597 scale

2235: 355.078 423.409 :T

2236: -355.078 -423.409 :T

2237: 314.078 427.409 :M

2238: f3_10 sf

2239: (TENSE      PRES)S

2240: gR

2241: gR

2242: gS 192 264 63 9 rC

2243: gS

2244: .611 .597 scale

2245: 365.714 449.227 :T

2246: -365.714 -449.227 :T

2247: 315.714 454.227 :M

2248: f3_10 sf

2249: (PRED)S

2250: gR

2251: gS

2252: .611 .597 scale

2253: 365.714 449.227 :T

2254: -365.714 -449.227 :T

2255: 341.887 454.227 :M

2256: f3_12 sf

2257: (      'eat\()S

2258: f3_10 sf

2259: (SUBJ)S

2260: f3_12 sf

2261: (\)')S

2262: gR

2263: gR

2264: gS 29 30 538 781 rC

2265: 161 298 :M

2266: f3_10 sf

2267: .822 .082(Figure 4)J

2268: 70 321 :M

2269: f3_11 sf

2270: 5.636 .564(Finally, Bod & Kaplan present a )J

2271: 268 321 :M

2272: -.193(third)A

2273: 70 335 :M

2274: 1.692 .169(decomposition operation, )J

2275: 193 335 :M

2276: f4_11 sf

2277: -.073(Discard)A

2278: 229 335 :M

2279: f3_11 sf

2280: 2.649 .265(, defined to)J

2281: 70 349 :M

2282: 3.024 .302(construct generalizations )J

2283: 197 349 :M

2284: 5.121 .512(of the fragments)J

2285: 70 363 :M

2286: .719 .072(supplied by )J

2287: 127 363 :M

2288: f4_11 sf

2289: .35(Root)A

2290: f3_11 sf

2291: .601 .06( and )J

2292: 173 363 :M

2293: f4_11 sf

2294: -.039(Frontier)A

2295: 211 363 :M

2296: f3_11 sf

2297: 1.364 .136(. )J

2298: 219 363 :M

2299: f4_11 sf

2300: .171(Discard)A

2301: f3_11 sf

2302: .306 .031( acts )J

2303: 280 363 :M

2304: -.556(to)A

2305: 70 377 :M

2306: 3.356 .336(delete combinations of attribute-value )J

2307: 267 377 :M

2308: -.094(pairs)A

2309: 70 391 :M

2310: .136 .014(subject to the )J

2311: 133 391 :M

2312: .28 .028(following condition: )J

2313: 228 391 :M

2314: f4_11 sf

2315: .057(Discard)A

2316: f3_11 sf

2317: .183 .018( does)J

2318: 70 406 :M

2319: 1.632 .163(not delete pairs whose values )J

2320: f1_11 sf

2321: (f)S

2322: 222 406 :M

2323: f3_11 sf

2324: .822 .082(-correspond )J

2325: 280 406 :M

2326: -.556(to)A

2327: 70 421 :M

2328: -.009(remaining c-structure nodes. According to Bod &)A

2329: 70 435 :M

2330: .196 .02(Kaplan \(1998\), )J

2331: 141 435 :M

2332: f4_11 sf

2333: -.073(Discard)A

2334: 177 435 :M

2335: f3_11 sf

2336: .394 .039(-generated fragments are)J

2337: 70 449 :M

2338: -.082(needed to parse sentences that are "ungrammatical)A

2339: 70 463 :M

2340: .9 .09(with respect to the )J

2341: 160 463 :M

2342: 1.091 .109(corpus", thus increasing the)J

2343: 70 477 :M

2344: -.023(robustness of the model.)A

2345: 70 500 :M

2346: f0_11 sf

2347: -.004(2.3  The composition operation)A

2348: 70 514 :M

2349: f3_11 sf

2350: 4.341 .434(In LFG-DOP the operation for )J

2351: 242 514 :M

2352: -.131(combining)A

2353: 70 528 :M

2354: -.081(fragments )A

2355: 117 528 :M

2356: .403 .04(is carried out in two steps. First the c)J

2357: 285 528 :M

2358: (-)S

2359: 70 542 :M

2360: -.05(structures )A

2361: 116 542 :M

2362: .634 .063(are combined by leftmost substitution)J

2363: 70 556 :M

2364: .136 .014(subject to the )J

2365: 133 556 :M

2366: .361 .036(category-matching condition, as in)J

2367: 70 570 :M

2368: 2.242 .224(Tree-DOP. This is followed by )J

2369: 229 570 :M

2370: 1.14 .114(the recursive)J

2371: 70 584 :M

2372: -.089(unification of the f-structures corresponding )A

2373: 264 584 :M

2374: -.149(to the)A

2375: 70 598 :M

2376: .881 .088(matching nodes. A derivation for an LFG-DOP)J

2377: 70 612 :M

2378: .085 .008(representation )J

2379: f4_11 sf

2380: (R)S

2381: 143 612 :M

2382: f3_11 sf

2383: 1.045 .104( is a sequence of fragments the)J

2384: 70 626 :M

2385: -.071(first of which )A

2386: 132 626 :M

2387: .023 .002(is labeled with )J

2388: 200 626 :M

2389: f4_11 sf

2390: .095(S)A

2391: f3_11 sf

2392: .243 .024( and for which the)J

2393: 70 640 :M

2394: -.096(iterative application of )A

2395: 171 640 :M

2396: .259 .026(the composition operation)J

2397: 70 654 :M

2398: -.017(produces )A

2399: f4_11 sf

2400: (R)S

2401: 120 654 :M

2402: f3_11 sf

2403: -.04(. For an illustration of the composition)A

2404: 70 668 :M

2405: -.053(operation, see Bod & Kaplan \(1998\).)A

2406: 70 691 :M

2407: f0_11 sf

2408: -.007(2.4  Probability models)A

2409: 70 705 :M

2410: f3_11 sf

2411: .501 .05(As in Tree-DOP, an LFG-DOP representation )J

2412: 281 705 :M

2413: f4_11 sf

2414: (R)S

2415: 70 719 :M

2416: f3_11 sf

2417: -.093(can typically be derived in many different ways. If)A

2418: 70 733 :M

2419: -.216(each derivation )A

2420: 138 733 :M

2421: f4_11 sf

2422: (D)S

2423: f3_11 sf

2424: -.024( has a probability P\()A

2425: 235 733 :M

2426: f4_11 sf

2427: (D)S

2428: 244 733 :M

2429: f3_11 sf

2430: -.029(\), then the)A

2431: 305 96 :M

2432: 1.635 .164(probability of deriving )J

2433: 417 96 :M

2434: f4_11 sf

2435: 2.505(R)A

2436: f3_11 sf

2437: 1.025 .103( )J

2438: 431 96 :M

2439: 3.267 .327(is the sum of the)J

2440: 305 110 :M

2441: -.164(individual derivation probabilities:)A

2442: 305 133 :M

2443: -.41(\(1\))A

2444: 340 133 :M

2445: .084(P\()A

2446: f4_11 sf

2447: .115(R)A

2448: f3_11 sf

2449: .079 .008(\)  =  )J

2450: 379 133 :M

2451: f1_14 sf

2452: .121(S)A

2453: f4_10 sf

2454: 0 2 rm

2455: .105(D)A

2456: 0 -2 rm

2457: f3_10 sf

2458: 0 2 rm

2459: .17 .017( derives )J

2460: 0 -2 rm

2461: f4_10 sf

2462: 0 2 rm

2463: .089(R)A

2464: 0 -2 rm

2465: f3_11 sf

2466: .152 .015( P\()J

2467: 449 133 :M

2468: f4_11 sf

2469: -.412(D)A

2470: f3_11 sf

2471: <29>S

2472: 305 154 :M

2473: 3.542 .354(An LFG-DOP )J

2474: 382 154 :M

2475: 3.222 .322(derivation is produced by a)J

2476: 305 168 :M

2477: 2.318 .232(stochastic process which )J

2478: 430 168 :M

2479: 3.009 .301(starts by randomly)J

2480: 305 182 :M

2481: .375 .037(choosing a fragment whose c-structure is )J

2482: 493 182 :M

2483: -.293(labeled)A

2484: 305 196 :M

2485: -.186(with the )A

2486: 343 196 :M

2487: -.013(initial category. At each subsequent step,)A

2488: 305 210 :M

2489: -.086(a next fragment is chosen )A

2490: 419 210 :M

2491: .591 .059(at random from among)J

2492: 305 224 :M

2493: 1.425 .143(the fragments that )J

2494: 395 224 :M

2495: 2.184 .218(can be composed with the)J

2496: 305 238 :M

2497: 1.974 .197(current subanalysis. )J

2498: 404 238 :M

2499: 3.152 .315(The chosen fragment is)J

2500: 305 252 :M

2501: -.059(composed with the current subanalysis )A

2502: 477 252 :M

2503: -.081(to produce)A

2504: 305 266 :M

2505: 1.144 .114(a new one; the process stops when an analysis)J

2506: 305 280 :M

2507: .296 .03(results with no non-terminal leaves. We )J

2508: 487 280 :M

2509: -.092(will call)A

2510: 305 294 :M

2511: .317 .032(the set of composable fragments at a certain )J

2512: 506 294 :M

2513: -.239(step)A

2514: 305 308 :M

2515: -.082(in the stochastic process the )A

2516: f4_11 sf

2517: -.088(competition\312set)A

2518: f3_11 sf

2519: -.084( at that)A

2520: 305 322 :M

2521: 1.332 .133(step. Let CP\()J

2522: 368 322 :M

2523: f4_11 sf

2524: (f)S

2525: 371 322 :M

2526: f3_11 sf

2527: 1.648 .165( | CS\) denote the probability of)J

2528: 305 336 :M

2529: -.11(choosing a )A

2530: 355 336 :M

2531: .446 .045(fragment )J

2532: f4_11 sf

2533: .097 .01(f )J

2534: f3_11 sf

2535: .361 .036(from a competition set CS)J

2536: 305 350 :M

2537: -.218(containing )A

2538: 353 350 :M

2539: f4_11 sf

2540: -.058(f, )A

2541: f3_11 sf

2542: -.088(then the probability of a derivation )A

2543: 515 350 :M

2544: f4_11 sf

2545: (D)S

2546: 305 364 :M

2547: f3_11 sf

2548: -.079(= <)A

2549: 321 364 :M

2550: f4_11 sf

2551: -.351(f)A

2552: f3_10 sf

2553: 0 2 rm

2554: -.574(1)A

2555: 0 -2 rm

2556: f3_11 sf

2557: -.631(, )A

2558: 334 364 :M

2559: f4_11 sf

2560: -.253(f)A

2561: f3_10 sf

2562: 0 2 rm

2563: -.414(2)A

2564: 0 -2 rm

2565: f3_11 sf

2566: -.285( ... )A

2567: 355 364 :M

2568: f4_11 sf

2569: (f)S

2570: 358 366 :M

2571: f4_10 sf

2572: .059(k)A

2573: f3_11 sf

2574: 0 -2 rm

2575: .165 .017(> is)J

2576: 0 2 rm

2577: 305 388 :M

2578: -.41(\(2\))A

2579: 340 388 :M

2580: (P\(<)S

2581: 357 388 :M

2582: f4_11 sf

2583: -.351(f)A

2584: f3_10 sf

2585: 0 2 rm

2586: -.574(1)A

2587: 0 -2 rm

2588: f3_11 sf

2589: -.631(, )A

2590: 370 388 :M

2591: f4_11 sf

2592: -.253(f)A

2593: f3_10 sf

2594: 0 2 rm

2595: -.414(2)A

2596: 0 -2 rm

2597: f3_11 sf

2598: -.285( ... )A

2599: 391 388 :M

2600: f4_11 sf

2601: (f)S

2602: 394 390 :M

2603: f4_10 sf

2604: .07(k)A

2605: f3_11 sf

2606: 0 -2 rm

2607: .092 .009(>\)  =  )J

2608: 0 2 rm

2609: 427 388 :M

2610: f1_14 sf

2611: -1.406(P)A

2612: f4_11 sf

2613: 0 2 rm

2614: (i)S

2615: 0 -2 rm

2616: 440 388 :M

2617: f3_10 sf

2618: -.074( )A

2619: f3_11 sf

2620: -.168(CP\()A

2621: f4_11 sf

2622: (f)S

2623: 463 390 :M

2624: f4_10 sf

2625: (i)S

2626: f3_11 sf

2627: 0 -2 rm

2628: -.051( | CS)A

2629: 0 2 rm

2630: f4_10 sf

2631: (i)S

2632: f3_11 sf

2633: 0 -2 rm

2634: <29>S

2635: 0 2 rm

2636: 305 409 :M

2637: .757 .076(where the )J

2638: 354 409 :M

2639: f4_11 sf

2640: -.162(competition\312probability)A

2641: 455 409 :M

2642: f3_11 sf

2643: 1.25 .125( )J

2644: 460 409 :M

2645: -.33(CP\()A

2646: f4_11 sf

2647: (f)S

2648: 480 409 :M

2649: f3_11 sf

2650: 2.133 .213( | CS\) is)J

2651: 305 423 :M

2652: -.054(expressed in terms of fragment probabilities P\()A

2653: 510 423 :M

2654: f4_11 sf

2655: -.554(f)A

2656: f3_11 sf

2657: -1.218(\):)A

2658: 408 446 10 19 rC

2659: gS

2660: .941 .854 scale

2661: 438.569 533.032 :T

2662: -438.569 -533.032 :T

2663: 434.569 540.032 :M

2664: f1_16 sf

2665: (S)S

2666: gR

2667: gR

2668: gS 416 454 7 11 rC

2669: gS

2670: .941 .854 scale

2671: 445.069 538.375 :T

2672: -445.069 -538.375 :T

2673: 443.069 542.375 :M

2674: f4_10 sf

2675: (f')S

2676: gR

2677: gR

2678: gS 421 453 8 12 rC

2679: gS

2680: .941 .854 scale

2681: 451.382 537.375 :T

2682: -451.382 -537.375 :T

2683: 448.382 542.375 :M

2684: f1_10 sf

2685: <CE>S

2686: gR

2687: gR

2688: gS 427 454 14 11 rC

2689: gS

2690: .941 .854 scale

2691: 460.757 538.375 :T

2692: -460.757 -538.375 :T

2693: 454.757 542.375 :M

2694: f3_10 sf

2695: (CS)S

2696: gR

2697: gR

2698: gS 440 451 15 12 rC

2699: gS

2700: .941 .854 scale

2701: 475.57 535.032 :T

2702: -475.57 -535.032 :T

2703: 468.57 540.032 :M

2704: f3_12 sf

2705: ( P\()S

2706: gR

2707: gR

2708: gS 453 451 7 12 rC

2709: gS

2710: .941 .854 scale

2711: 485.382 535.032 :T

2712: -485.382 -535.032 :T

2713: 482.382 540.032 :M

2714: f4_12 sf

2715: (f')S

2716: gR

2717: gR

2718: gS 458 451 6 12 rC

2719: gS

2720: .941 .854 scale

2721: 489.695 535.032 :T

2722: -489.695 -535.032 :T

2723: 487.695 540.032 :M

2724: f3_12 sf

2725: <29>S

2726: gR

2727: gR

2728: gS 426 432 13 13 rC

2729: gS

2730: .941 .854 scale

2731: 458.694 512.775 :T

2732: -458.694 -512.775 :T

2733: 453.694 517.775 :M

2734: f3_12 sf

2735: (P\()S

2736: gR

2737: gR

2738: gS 437 432 4 13 rC

2739: gS

2740: .941 .854 scale

2741: 466.382 512.775 :T

2742: -466.382 -512.775 :T

2743: 465.382 517.775 :M

2744: f4_12 sf

2745: (f)S

2746: gR

2747: gR

2748: gS 440 432 5 13 rC

2749: gS

2750: .941 .854 scale

2751: 470.57 512.775 :T

2752: -470.57 -512.775 :T

2753: 468.57 517.775 :M

2754: f3_12 sf

2755: <29>S

2756: gR

2757: gR

2758: gS 443 432 5 13 rC

2759: gS

2760: .941 .854 scale

2761: 472.757 512.775 :T

2762: -472.757 -512.775 :T

2763: 471.757 517.775 :M

2764: f3_12 sf

2765: ( )S

2766: gR

2767: gR

2768: gS 446 439 3 4 rC

2769: gS

2770: .941 .854 scale

2771: 474.945 515.775 :T

2772: -474.945 -515.775 :T

2773: 474.945 517.775 :M

2774: f3_4 sf

2775: ( )S

2776: gR

2777: gR

2778: gS 305 431 160 35 rC

2779: 408 448 -1 1 463 447 1 408 447 @a

2780: 306 440 15 12 rC

2781: gS

2782: .941 .854 scale

2783: 333.192 522.146 :T

2784: -333.192 -522.146 :T

2785: 326.192 527.146 :M

2786: f3_12 sf

2787: (\(3\))S

2788: gR

2789: gR

2790: gS 339 440 20 12 rC

2791: gS

2792: .941 .854 scale

2793: 370.256 522.146 :T

2794: -370.256 -522.146 :T

2795: 361.256 527.146 :M

2796: f3_12 sf

2797: (CP\()S

2798: gR

2799: gR

2800: gS 357 440 4 12 rC

2801: gS

2802: .941 .854 scale

2803: 381.381 522.146 :T

2804: -381.381 -522.146 :T

2805: 380.381 527.146 :M

2806: f4_12 sf

2807: (f)S

2808: gR

2809: gR

2810: gS 360 440 39 12 rC

2811: gS

2812: .941 .854 scale

2813: 402.568 522.146 :T

2814: -402.568 -522.146 :T

2815: 383.568 527.146 :M

2816: f3_12 sf

2817: ( | CS\)  =)S

2818: gR

2819: gR

2820: gS 29 30 538 781 rC

2821: 305 482 :M

2822: f3_11 sf

2823: -.101(Bod & Kaplan give three )A

2824: 417 482 :M

2825: -.038(definitions of increasing)A

2826: 305 496 :M

2827: 2.479 .248(complexity for the competition set: the first)J

2828: 305 510 :M

2829: .707 .071(definition groups all fragments that )J

2830: 470 510 :M

2831: .944 .094(only satisfy)J

2832: 305 524 :M

2833: 5.221 .522(the Category-matching condition )J

2834: 486 524 :M

2835: 7.767 .777(of the)J

2836: 305 538 :M

2837: 2.267 .227(composition operation; )J

2838: 420 538 :M

2839: 2.688 .269(the second definition)J

2840: 305 552 :M

2841: -.061(groups all fragments which satisfy both Category)A

2842: 520 552 :M

2843: (-)S

2844: 305 566 :M

2845: -.074(matching and Uniqueness; and )A

2846: 442 566 :M

2847: -.078(the third definition)A

2848: 305 580 :M

2849: 2.168 .217(groups all fragments )J

2850: 410 580 :M

2851: 1.852 .185(which satisfy Category)J

2852: 520 580 :M

2853: (-)S

2854: 305 594 :M

2855: 1.125 .112(matching, Uniqueness )J

2856: 411 594 :M

2857: 2.081 .208(and Coherence. Bod &)J

2858: 305 608 :M

2859: -.252(Kaplan )A

2860: 339 608 :M

2861: .043 .004(point out that the Completeness condition)J

2862: 305 622 :M

2863: .146 .015(cannot be enforced at )J

2864: 404 622 :M

2865: .631 .063(each step of the stochastic)J

2866: 305 636 :M

2867: .221 .022(derivation process, and is )J

2868: 422 636 :M

2869: .662 .066(a property of the final)J

2870: 305 650 :M

2871: .831 .083(representation which )J

2872: 404 650 :M

2873: 2.15 .215(can only be enforced by)J

2874: 305 664 :M

2875: .048 .005(sampling valid representations from the output of)J

2876: 305 678 :M

2877: -.165(the stochastic )A

2878: 366 678 :M

2879: .321 .032(process. In this paper, we will only)J

2880: 305 692 :M

2881: -.129(deal with the third definition of competition )A

2882: 496 692 :M

2883: .077 .008(set, as)J

2884: 305 706 :M

2885: .256 .026(it selects only those fragments )J

2886: 444 706 :M

2887: -.011(at each derivation)A

2888: 305 720 :M

2889: 1.535 .153(step that may )J

2890: 375 720 :M

2891: 1.917 .192(finally result into a valid LFG)J

2892: 305 734 :M

2893: .136 .014(representation, thus reducing the off-line validity)J

2894: 305 748 :M

2895: -.098(checking to the Completeness condition.)A

2896: endp

2897: %%Page: 4 4

2898: %%BeginPageSetup

2899: initializepage

2900: (rens; page: 4 of 8)setjob

2901: %%EndPageSetup

2902: -29 -30 :T

2903: gS 29 30 538 781 rC

2904: 105 96 :M

2905: f3_11 sf

2906: 4.907 .491(Note )J

2907: 137 96 :M

2908: 4.921 .492(that the computation of the)J

2909: 70 110 :M

2910: -.063(competition probability in the above formulas still)A

2911: 70 124 :M

2912: -.015(requires a definition )A

2913: 162 124 :M

2914: .443 .044(for the fragment probability)J

2915: 70 138 :M

2916: -.317(P\()A

2917: f4_11 sf

2918: (f)S

2919: 83 138 :M

2920: f3_11 sf

2921: .514 .051(\). Bod and Kaplan define the probability of a)J

2922: 70 152 :M

2923: .828 .083(fragment simply as its relative frequency )J

2924: 262 152 :M

2925: .837 .084(in the)J

2926: 70 166 :M

2927: .699 .07(bag of all fragments generated from the corpus,)J

2928: 70 180 :M

2929: .609 .061(just as in most Tree-DOP models. We will )J

2930: 268 180 :M

2931: -.186(refer)A

2932: 70 194 :M

2933: 1.95 .195(to this fragment estimator as "simple relative)J

2934: 70 208 :M

2935: -.015(frequency" or "simple RF".)A

2936: 105 222 :M

2937: -.013(\027We will also use an alternative definition)A

2938: 70 236 :M

2939: .233 .023(of fragment probability which is )J

2940: 218 236 :M

2941: .536 .054(a refinement of)J

2942: 70 250 :M

2943: .39 .039(simple RF. This alternative fragment probability)J

2944: 70 264 :M

2945: 3.741 .374(definition )J

2946: 124 264 :M

2947: 3.888 .389(distinguishes between fragments)J

2948: 70 278 :M

2949: 3.876 .388(supplied )J

2950: 118 278 :M

2951: 6.875 .688(by )J

2952: 142 278 :M

2953: f4_11 sf

2954: -.181(Root)A

2955: f3_11 sf

2956: (/)S

2957: 166 278 :M

2958: f4_11 sf

2959: -.039(Frontier)A

2960: 204 278 :M

2961: f3_11 sf

2962: 5.649 .565( and fragments)J

2963: 70 292 :M

2964: .149 .015(supplied by )J

2965: f4_11 sf

2966: .052(Discard)A

2967: 161 292 :M

2968: f3_11 sf

2969: .417 .042(. We will )J

2970: 207 292 :M

2971: .434 .043(treat the first type)J

2972: 70 306 :M

2973: -.455(of )A

2974: 82 306 :M

2975: .3 .03(fragments as seen events, and the second type)J

2976: 70 320 :M

2977: .908 .091(of )J

2978: 84 320 :M

2979: 1.447 .145(fragments as previously unseen events. We)J

2980: 70 334 :M

2981: .654 .065(thus )J

2982: 93 334 :M

2983: 1.205 .121(create two separate bags corresponding to)J

2984: 70 348 :M

2985: (two separate )S

2986: 129 348 :M

2987: .629 .063(distributions: a bag with fragments)J

2988: 70 362 :M

2989: -.02(generated by )A

2990: 130 362 :M

2991: f4_11 sf

2992: .104(Root)A

2993: f3_11 sf

2994: .178 .018( and )J

2995: 174 362 :M

2996: f4_11 sf

2997: -.039(Frontier)A

2998: 212 362 :M

2999: f3_11 sf

3000: 1.1 .11(, and a bag with)J

3001: 70 376 :M

3002: 2.505 .251(fragments generated by )J

3003: 190 376 :M

3004: f4_11 sf

3005: .677(Discard)A

3006: f3_11 sf

3007: 2.273 .227(. We assign)J

3008: 70 390 :M

3009: .203 .02(probability mass to the fragments of each bag by)J

3010: 70 404 :M

3011: .443 .044(means of )J

3012: 115 404 :M

3013: f4_11 sf

3014: -.133(discounting)A

3015: 166 404 :M

3016: f3_11 sf

3017: (: the )S

3018: 189 404 :M

3019: .162 .016(relative frequencies of)J

3020: 70 418 :M

3021: 3.485 .349(seen events are )J

3022: 156 418 :M

3023: 3.232 .323(discounted and the gained)J

3024: 70 432 :M

3025: -.014(probability mass is reserved for the bag of unseen)A

3026: 70 446 :M

3027: .399 .04(events \(cf. Ney et al. 1997\). We )J

3028: 218 446 :M

3029: .111 .011(accomplish this)J

3030: 70 460 :M

3031: 2.343 .234(by a very simple estimator: the )J

3032: 231 460 :M

3033: -.104(Turing-Good)A

3034: 70 474 :M

3035: 2.42 .242(estimator \(Good )J

3036: 154 474 :M

3037: 3.135 .314(1953\) which computes the)J

3038: 70 488 :M

3039: .308 .031(probability mass of unseen )J

3040: 194 488 :M

3041: .328 .033(events as\312)J

3042: f4_11 sf

3043: (n)S

3044: 244 490 :M

3045: f3_10 sf

3046: -.656(1)A

3047: f3_11 sf

3048: 0 -2 rm

3049: (/)S

3050: 0 2 rm

3051: 252 488 :M

3052: f4_11 sf

3053: -.189(N)A

3054: f3_11 sf

3055: -.153(\312where)A

3056: 70 503 :M

3057: f4_11 sf

3058: (n)S

3059: 76 505 :M

3060: f3_10 sf

3061: .062(1)A

3062: f3_11 sf

3063: 0 -2 rm

3064: .178 .018( is the number of singleton events and )J

3065: 0 2 rm

3066: 254 503 :M

3067: f4_11 sf

3068: .085(N)A

3069: f3_11 sf

3070: .117 .012( is the)J

3071: 70 518 :M

3072: 2.25 .225(total number of seen events. )J

3073: 215 518 :M

3074: .972 .097(This probability)J

3075: 70 532 :M

3076: .063 .006(mass is )J

3077: 106 532 :M

3078: .324 .032(assigned to the bag of )J

3079: 208 532 :M

3080: f4_11 sf

3081: -.073(Discard)A

3082: 244 532 :M

3083: f3_11 sf

3084: -.267(-generated)A

3085: 70 547 :M

3086: 1.913 .191(fragments. The remaining )J

3087: 198 547 :M

3088: 2.532 .253(mass \(1\312)J

3089: f1_11 sf

3090: (-)S

3091: 247 547 :M

3092: f3_11 sf

3093: -.417<CA>A

3094: f4_11 sf

3095: (n)S

3096: 255 549 :M

3097: f3_10 sf

3098: -.656(1)A

3099: f3_11 sf

3100: 0 -2 rm

3101: (/)S

3102: 0 2 rm

3103: 263 547 :M

3104: f4_11 sf

3105: 1.364(N)A

3106: f3_11 sf

3107: 1.965 .197(\) is)J

3108: 70 562 :M

3109: 1.713 .171(assigned to the bag of )J

3110: 182 562 :M

3111: f4_11 sf

3112: -.181(Root)A

3113: f3_11 sf

3114: (/)S

3115: 206 562 :M

3116: f4_11 sf

3117: -.039(Frontier)A

3118: 244 562 :M

3119: f3_11 sf

3120: -.267(-generated)A

3121: 70 576 :M

3122: .762 .076(fragments. )J

3123: 122 576 :M

3124: 1.374 .137(The probability of each fragment is)J

3125: 70 590 :M

3126: .106 .011(then computed as its relative frequency in its bag)J

3127: 70 604 :M

3128: -.078(multiplied by the probability mass assigned to )A

3129: 273 604 :M

3130: -.297(this)A

3131: 70 618 :M

3132: -.204(bag. Let |\312)A

3133: f4_11 sf

3134: -.298<66CA>A

3135: 118 618 :M

3136: f3_11 sf

3137: .102 .01(| denote the frequency of a fragment )J

3138: 283 618 :M

3139: f4_11 sf

3140: -.951(f)A

3141: f3_11 sf

3142: (,)S

3143: 70 632 :M

3144: -.099(then its probability is given by:)A

3145: 214 674 6 13 rC

3146: gS

3147: .899 .886 scale

3148: 241.211 767.26 :T

3149: -241.211 -767.26 :T

3150: 239.211 772.26 :M

3151: f3_12 sf

3152: <7CCA>S

3153: gR

3154: gR

3155: gS 218 674 7 13 rC

3156: gS

3157: .899 .886 scale

3158: 246.661 767.26 :T

3159: -246.661 -767.26 :T

3160: 243.661 772.26 :M

3161: f4_12 sf

3162: <66CA>S

3163: gR

3164: gR

3165: gS 224 674 3 13 rC

3166: gS

3167: .899 .886 scale

3168: 251.337 767.26 :T

3169: -251.337 -767.26 :T

3170: 250.337 772.26 :M

3171: f3_12 sf

3172: (|)S

3173: gR

3174: gR

3175: gS 153 690 9 18 rC

3176: gS

3177: .899 .886 scale

3178: 174.342 789.841 :T

3179: -174.342 -789.841 :T

3180: 171.342 794.841 :M

3181: f1_14 sf

3182: (S)S

3183: gR

3184: gR

3185: gS 160 697 6 11 rC

3186: gS

3187: .899 .886 scale

3188: 181.13 792.97 :T

3189: -181.13 -792.97 :T

3190: 179.13 795.97 :M

3191: f4_9 sf

3192: (f')S

3193: gR

3194: gR

3195: gS 163 697 7 11 rC

3196: gS

3197: .899 .886 scale

3198: 184.468 792.97 :T

3199: -184.468 -792.97 :T

3200: 182.468 795.97 :M

3201: f3_9 sf

3202: (: )S

3203: gR

3204: gR

3205: gS 169 697 6 11 rC

3206: gS

3207: .899 .886 scale

3208: 191.144 792.97 :T

3209: -191.144 -792.97 :T

3210: 189.144 795.97 :M

3211: f4_9 sf

3212: (f')S

3213: gR

3214: gR

3215: gS 172 697 58 11 rC

3216: gS

3217: .899 .886 scale

3218: 222.481 792.97 :T

3219: -222.481 -792.97 :T

3220: 192.481 795.97 :M

3221: f3_9 sf

3222: -.006( is generated by )A

3223: gR

3224: gR

3225: gS 229 697 18 11 rC

3226: gS

3227: .899 .886 scale

3228: 264.9 792.97 :T

3229: -264.9 -792.97 :T

3230: 255.9 795.97 :M

3231: f4_9 sf

3232: (Root)S

3233: gR

3234: gR

3235: gS 244 697 4 11 rC

3236: gS

3237: .899 .886 scale

3238: 272.589 792.97 :T

3239: -272.589 -792.97 :T

3240: 272.589 795.97 :M

3241: f3_9 sf

3242: (/)S

3243: gR

3244: gR

3245: gS 247 697 31 11 rC

3246: gS

3247: .899 .886 scale

3248: 291.927 792.97 :T

3249: -291.927 -792.97 :T

3250: 275.927 795.97 :M

3251: f4_9 sf

3252: -.008(Frontier)A

3253: gR

3254: gR

3255: gS 275 697 4 11 rC

3256: gS

3257: .899 .886 scale

3258: 308.08 791.97 :T

3259: -308.08 -791.97 :T

3260: 307.08 795.97 :M

3261: f3_10 sf

3262: ( )S

3263: gR

3264: gR

3265: gS 278 693 6 13 rC

3266: gS

3267: .899 .886 scale

3268: 312.418 789.841 :T

3269: -312.418 -789.841 :T

3270: 310.418 794.841 :M

3271: f3_12 sf

3272: <7CCA>S

3273: gR

3274: gR

3275: gS 282 693 7 13 rC

3276: gS

3277: .899 .886 scale

3278: 316.868 789.841 :T

3279: -316.868 -789.841 :T

3280: 314.868 794.841 :M

3281: f4_12 sf

3282: (f')S

3283: gR

3284: gR

3285: gS 288 693 3 13 rC

3286: gS

3287: .899 .886 scale

3288: 320.431 789.841 :T

3289: -320.431 -789.841 :T

3290: 320.431 794.841 :M

3291: f3_12 sf

3292: (|)S

3293: gR

3294: gR

3295: gS 70 647 222 62 rC

3296: 154 691 -1 1 289 690 1 154 690 @a

3297: 104 682 14 14 rC

3298: gS

3299: .899 .886 scale

3300: 122.824 777.422 :T

3301: -122.824 -777.422 :T

3302: 116.824 782.422 :M

3303: f3_12 sf

3304: (\(1 )S

3305: gR

3306: gR

3307: gS 116 682 8 15 rC

3308: gS

3309: .899 .886 scale

3310: 133.175 777.422 :T

3311: -133.175 -777.422 :T

3312: 130.175 782.422 :M

3313: f1_12 sf

3314: (-)S

3315: gR

3316: gR

3317: gS 122 682 5 14 rC

3318: gS

3319: .899 .886 scale

3320: 137.851 777.422 :T

3321: -137.851 -777.422 :T

3322: 136.851 782.422 :M

3323: f3_12 sf

3324: ( )S

3325: gR

3326: gR

3327: gS 125 682 7 14 rC

3328: gS

3329: .899 .886 scale

3330: 143.189 777.422 :T

3331: -143.189 -777.422 :T

3332: 140.189 782.422 :M

3333: f4_12 sf

3334: (n)S

3335: gR

3336: gR

3337: gS 130 686 7 11 rC

3338: gS

3339: .899 .886 scale

3340: 147.752 780.68 :T

3341: -147.752 -780.68 :T

3342: 145.752 784.68 :M

3343: f3_10 sf

3344: (1)S

3345: gR

3346: gR

3347: gS 135 682 4 14 rC

3348: gS

3349: .899 .886 scale

3350: 152.315 777.422 :T

3351: -152.315 -777.422 :T

3352: 151.315 782.422 :M

3353: f3_12 sf

3354: (/)S

3355: gR

3356: gR

3357: gS 137 682 9 14 rC

3358: gS

3359: .899 .886 scale

3360: 157.54 777.422 :T

3361: -157.54 -777.422 :T

3362: 153.54 782.422 :M

3363: f4_12 sf

3364: (N)S

3365: gR

3366: gR

3367: gS 145 682 5 14 rC

3368: gS

3369: .899 .886 scale

3370: 163.328 777.422 :T

3371: -163.328 -777.422 :T

3372: 161.328 782.422 :M

3373: f3_12 sf

3374: <29>S

3375: gR

3376: gR

3377: gS 71 648 14 13 rC

3378: gS

3379: .899 .886 scale

3380: 87.108 739.034 :T

3381: -87.108 -739.034 :T

3382: 80.108 744.034 :M

3383: f3_12 sf

3384: (\(4\))S

3385: gR

3386: gR

3387: gS 102 648 12 13 rC

3388: gS

3389: .899 .886 scale

3390: 119.599 739.034 :T

3391: -119.599 -739.034 :T

3392: 114.599 744.034 :M

3393: f3_12 sf

3394: (P\()S

3395: gR

3396: gR

3397: gS 112 648 5 13 rC

3398: gS

3399: .899 .886 scale

3400: 126.725 739.034 :T

3401: -126.725 -739.034 :T

3402: 125.725 744.034 :M

3403: f4_12 sf

3404: (f)S

3405: gR

3406: gR

3407: gS 115 648 9 13 rC

3408: gS

3409: .899 .886 scale

3410: 132.063 739.034 :T

3411: -132.063 -739.034 :T

3412: 129.063 744.034 :M

3413: f3_12 sf

3414: ( | )S

3415: gR

3416: gR

3417: gS 122 648 5 13 rC

3418: gS

3419: .899 .886 scale

3420: 137.851 739.034 :T

3421: -137.851 -739.034 :T

3422: 136.851 744.034 :M

3423: f4_12 sf

3424: (f)S

3425: gR

3426: gR

3427: gS 125 648 71 13 rC

3428: gS

3429: .899 .886 scale

3430: 178.189 739.034 :T

3431: -178.189 -739.034 :T

3432: 140.189 744.034 :M

3433: f3_12 sf

3434: ( is generated by )S

3435: gR

3436: gR

3437: gS 194 648 23 13 rC

3438: gS

3439: .899 .886 scale

3440: 227.959 739.034 :T

3441: -227.959 -739.034 :T

3442: 216.959 744.034 :M

3443: f4_12 sf

3444: (Root)S

3445: gR

3446: gR

3447: gS 215 648 4 13 rC

3448: gS

3449: .899 .886 scale

3450: 241.324 739.034 :T

3451: -241.324 -739.034 :T

3452: 240.324 744.034 :M

3453: f3_12 sf

3454: (/)S

3455: gR

3456: gR

3457: gS 217 648 39 13 rC

3458: gS

3459: .899 .886 scale

3460: 262.549 739.034 :T

3461: -262.549 -739.034 :T

3462: 242.549 744.034 :M

3463: f4_12 sf

3464: (Frontier)S

3465: gR

3466: gR

3467: gS 254 648 17 13 rC

3468: gS

3469: .899 .886 scale

3470: 291.715 739.034 :T

3471: -291.715 -739.034 :T

3472: 283.715 744.034 :M

3473: f3_12 sf

3474: (\)  =)S

3475: gR

3476: gR

3477: gS 306 86 15 14 rC

3478: gS

3479: .918 .914 scale

3480: 341.467 101.095 :T

3481: -341.467 -101.095 :T

3482: 334.467 106.095 :M

3483: f3_12 sf

3484: (\(5\))S

3485: gR

3486: gR

3487: gS 338 86 12 14 rC

3488: gS

3489: .918 .914 scale

3490: 374.33 101.095 :T

3491: -374.33 -101.095 :T

3492: 369.33 106.095 :M

3493: f3_12 sf

3494: (P\()S

3495: gR

3496: gR

3497: gS 348 86 5 14 rC

3498: gS

3499: .918 .914 scale

3500: 381.225 101.095 :T

3501: -381.225 -101.095 :T

3502: 380.225 106.095 :M

3503: f4_12 sf

3504: (f)S

3505: gR

3506: gR

3507: gS 351 86 9 14 rC

3508: gS

3509: .918 .914 scale

3510: 386.494 101.095 :T

3511: -386.494 -101.095 :T

3512: 383.494 106.095 :M

3513: f3_12 sf

3514: ( | )S

3515: gR

3516: gR

3517: gS 358 86 5 14 rC

3518: gS

3519: .918 .914 scale

3520: 392.12 101.095 :T

3521: -392.12 -101.095 :T

3522: 391.12 106.095 :M

3523: f4_12 sf

3524: (f)S

3525: gR

3526: gR

3527: gS 361 86 73 14 rC

3528: gS

3529: .918 .914 scale

3530: 432.388 101.095 :T

3531: -432.388 -101.095 :T

3532: 394.388 106.095 :M

3533: f3_12 sf

3534: ( is generated by )S

3535: gR

3536: gR

3537: gS 432 86 37 14 rC

3538: gS

3539: .918 .914 scale

3540: 490.741 101.095 :T

3541: -490.741 -101.095 :T

3542: 471.741 106.095 :M

3543: f4_12 sf

3544: (Discard)S

3545: gR

3546: gR

3547: gS 467 86 18 14 rC

3548: gS

3549: .918 .914 scale

3550: 517.872 101.095 :T

3551: -517.872 -101.095 :T

3552: 509.872 106.095 :M

3553: f3_12 sf

3554: (\)  =)S

3555: gR

3556: gR

3557: gS 341 121 5 13 rC

3558: gS

3559: .918 .914 scale

3560: 374.599 139.377 :T

3561: -374.599 -139.377 :T

3562: 372.599 144.377 :M

3563: f3_12 sf

3564: <28>S

3565: gR

3566: gR

3567: gS 344 121 8 13 rC

3568: gS

3569: .918 .914 scale

3570: 378.867 139.377 :T

3571: -378.867 -139.377 :T

3572: 375.867 144.377 :M

3573: f4_12 sf

3574: (n)S

3575: gR

3576: gR

3577: gS 350 124 6 12 rC

3578: gS

3579: .918 .914 scale

3580: 384.404 141.47 :T

3581: -384.404 -141.47 :T

3582: 382.404 145.47 :M

3583: f3_10 sf

3584: (1)S

3585: gR

3586: gR

3587: gS 355 121 4 13 rC

3588: gS

3589: .918 .914 scale

3590: 387.762 139.377 :T

3591: -387.762 -139.377 :T

3592: 386.762 144.377 :M

3593: f3_12 sf

3594: (/)S

3595: gR

3596: gR

3597: gS 357 121 9 13 rC

3598: gS

3599: .918 .914 scale

3600: 394.03 139.377 :T

3601: -394.03 -139.377 :T

3602: 390.03 144.377 :M

3603: f4_12 sf

3604: (N)S

3605: gR

3606: gR

3607: gS 365 121 5 13 rC

3608: gS

3609: .918 .914 scale

3610: 400.746 139.377 :T

3611: -400.746 -139.377 :T

3612: 398.746 144.377 :M

3613: f3_12 sf

3614: <29>S

3615: gR

3616: gR

3617: gS 424 112 7 14 rC

3618: gS

3619: .918 .914 scale

3620: 465.025 129.533 :T

3621: -465.025 -129.533 :T

3622: 463.025 134.533 :M

3623: f3_12 sf

3624: <7CCA>S

3625: gR

3626: gR

3627: gS 429 112 7 14 rC

3628: gS

3629: .918 .914 scale

3630: 471.472 129.533 :T

3631: -471.472 -129.533 :T

3632: 468.472 134.533 :M

3633: f4_12 sf

3634: <66CA>S

3635: gR

3636: gR

3637: gS 434 112 4 14 rC

3638: gS

3639: .918 .914 scale

3640: 474.92 129.533 :T

3641: -474.92 -129.533 :T

3642: 473.92 134.533 :M

3643: f3_12 sf

3644: (|)S

3645: gR

3646: gR

3647: gS 372 130 9 18 rC

3648: gS

3649: .918 .914 scale

3650: 410.372 152.502 :T

3651: -410.372 -152.502 :T

3652: 406.372 157.502 :M

3653: f1_14 sf

3654: (S)S

3655: gR

3656: gR

3657: gS 379 137 7 11 rC

3658: gS

3659: .918 .914 scale

3660: 415.999 155.595 :T

3661: -415.999 -155.595 :T

3662: 413.999 158.595 :M

3663: f4_9 sf

3664: (f')S

3665: gR

3666: gR

3667: gS 383 137 6 11 rC

3668: gS

3669: .918 .914 scale

3670: 420.357 155.595 :T

3671: -420.357 -155.595 :T

3672: 418.357 158.595 :M

3673: f3_9 sf

3674: (: )S

3675: gR

3676: gR

3677: gS 389 137 6 11 rC

3678: gS

3679: .918 .914 scale

3680: 425.804 155.595 :T

3681: -425.804 -155.595 :T

3682: 423.804 158.595 :M

3683: f4_9 sf

3684: (f')S

3685: gR

3686: gR

3687: gS 392 137 59 11 rC

3688: gS

3689: .918 .914 scale

3690: 459.162 155.595 :T

3691: -459.162 -155.595 :T

3692: 428.162 158.595 :M

3693: f3_9 sf

3694: -.006( is generated by )A

3695: gR

3696: gR

3697: gS 450 137 29 11 rC

3698: gS

3699: .918 .914 scale

3700: 505.351 155.595 :T

3701: -505.351 -155.595 :T

3702: 491.351 158.595 :M

3703: f4_9 sf

3704: -.008(Discard)A

3705: gR

3706: gR

3707: gS 478 136 4 12 rC

3708: gS

3709: .918 .914 scale

3710: 521.767 154.595 :T

3711: -521.767 -154.595 :T

3712: 520.767 158.595 :M

3713: f3_10 sf

3714: ( )S

3715: gR

3716: gR

3717: gS 480 133 7 13 rC

3718: gS

3719: .918 .914 scale

3720: 526.035 152.502 :T

3721: -526.035 -152.502 :T

3722: 524.035 157.502 :M

3723: f3_12 sf

3724: <7CCA>S

3725: gR

3726: gR

3727: gS 485 133 7 13 rC

3728: gS

3729: .918 .914 scale

3730: 531.483 152.502 :T

3731: -531.483 -152.502 :T

3732: 529.483 157.502 :M

3733: f4_12 sf

3734: (f')S

3735: gR

3736: gR

3737: gS 490 133 4 13 rC

3738: gS

3739: .918 .914 scale

3740: 534.93 152.502 :T

3741: -534.93 -152.502 :T

3742: 534.93 157.502 :M

3743: f3_12 sf

3744: (|)S

3745: gR

3746: gR

3747: gS 305 85 190 64 rC

3748: 374 130 -1 1 493 129 1 374 129 @a

3749: gR

3750: gS 29 30 538 781 rC

3751: 305 165 :M

3752: f3_11 sf

3753: 3.867 .387(We will refer )J

3754: 384 165 :M

3755: 3.042 .304(to this fragment probability)J

3756: 305 179 :M

3757: 1.19 .119(estimator as "discounted relative frequency" or)J

3758: 305 193 :M

3759: -.039("discounted RF".)A

3760: 305 217 :M

3761: f0_12 sf

3762: .649 .065(4  Parsing with LFG-DOP)J

3763: 305 231 :M

3764: f3_11 sf

3765: .908 .091(In )J

3766: 319 231 :M

3767: 1.647 .165(his PhD-thesis, Cormons \(1999\) presents a)J

3768: 305 245 :M

3769: .522 .052(parsing algorithm for LFG-DOP )J

3770: 455 245 :M

3771: .68 .068(which is based)J

3772: 305 259 :M

3773: .335 .034(on the Tree-DOP )J

3774: 386 259 :M

3775: .325 .032(parsing technique described in)J

3776: 305 273 :M

3777: 3.779 .378(Bod )J

3778: 332 273 :M

3779: 3.815 .382(\(1998\). Cormons first converts LFG)J

3780: 520 273 :M

3781: (-)S

3782: 305 287 :M

3783: .11 .011(representations into more compact indexed trees:)J

3784: 305 301 :M

3785: .178 .018(each node in the c-structure is )J

3786: 442 301 :M

3787: .5 .05(assigned an index)J

3788: 305 316 :M

3789: 1.373 .137(which refers to the )J

3790: 400 316 :M

3791: f1_11 sf

3792: .102(f)A

3793: f3_11 sf

3794: .583 .058(-corresponding f-structure)J

3795: 305 331 :M

3796: -.098(unit. For example, the representation in )A

3797: 478 331 :M

3798: .135 .014(figure 1 is)J

3799: 305 345 :M

3800: -.192(indexed as)A

3801: 305 368 :M

3802: (\(S.1)S

3803: 340 368 :M

3804: (\(NP.2)S

3805: 375 368 :M

3806: .107(Kim.2\))A

3807: 340 382 :M

3808: (\(VP.1)S

3809: 375 382 :M

3810: -.096(eats.1\)\))A

3811: 305 405 :M

3812: -.217(where)A

3813: 305 428 :M

3814: .143 .014(1 --> [  \(SUBJ = 2\))J

3815: 340 442 :M

3816: .75 .075(\(TENSE = PRES\))J

3817: 340 456 :M

3818: .13 .013(\(PRED = eat\(SUBJ\)\) ])J

3819: 305 479 :M

3820: .203 .02(2 --> [  \(PRED = Kim\))J

3821: 305 493 :M

3822: .107 .011(            \(NUM = SG\) ])J

3823: 305 516 :M

3824: 3.206 .321(The indexed trees )J

3825: 402 516 :M

3826: 3.589 .359(are then fragmented by)J

3827: 305 530 :M

3828: -.203(applying the )A

3829: 361 530 :M

3830: .022 .002(Tree-DOP decomposition operations)J

3831: 305 544 :M

3832: 1.11 .111(described )J

3833: 353 544 :M

3834: 2.911 .291(in section 2. Next, the LFG-DOP)J

3835: 305 558 :M

3836: 1.572 .157(decomposition operations )J

3837: 430 558 :M

3838: f4_11 sf

3839: .364(Root)A

3840: f3_11 sf

3841: .321 .032(, )J

3842: f4_11 sf

3843: .373(Frontier)A

3844: 499 558 :M

3845: f3_11 sf

3846: 3.36 .336( and)J

3847: 305 572 :M

3848: f4_11 sf

3849: -.073(Discard)A

3850: 341 572 :M

3851: f3_11 sf

3852: .844 .084( are applied to )J

3853: 412 572 :M

3854: 1.024 .102(the f-structure units that)J

3855: 305 586 :M

3856: 1.693 .169(correspond )J

3857: 361 586 :M

3858: 3.059 .306(to the indices in the c-structure)J

3859: 305 600 :M

3860: .789 .079(subtrees. Having obtained the set )J

3861: 461 600 :M

3862: 1.924 .192(of LFG-DOP)J

3863: 305 614 :M

3864: -.056(fragments in this way, each test sentence is parsed)A

3865: 305 628 :M

3866: 1.428 .143(by a bottom-up chart parser using initially )J

3867: 511 628 :M

3868: -.719(the)A

3869: 305 642 :M

3870: -.079(indexed subtrees only.)A

3871: 340 656 :M

3872: 8.037 .804(Thus )J

3873: 377 656 :M

3874: 5.021 .502(only the Category-matching)J

3875: 305 670 :M

3876: 1.902 .19(condition is enforced during )J

3877: 446 670 :M

3878: 1.074 .107(the chart-parsing)J

3879: 305 684 :M

3880: 5.833 .583(process. The )J

3881: 383 684 :M

3882: 4.482 .448(Uniqueness and Coherence)J

3883: 305 698 :M

3884: -.058(conditions )A

3885: 354 698 :M

3886: .558 .056(of the corresponding f-structure units)J

3887: 305 712 :M

3888: .029 .003(are enforced during )J

3889: 395 712 :M

3890: .419 .042(the disambiguation or chart)J

3891: 520 712 :M

3892: (-)S

3893: 305 726 :M

3894: 9.851 .985(decoding process. )J

3895: 424 726 :M

3896: 7.174 .717(Disambiguation is)J

3897: 305 740 :M

3898: .578 .058(accomplished by )J

3899: 385 740 :M

3900: 1.565 .156(computing a large number of)J

3901: endp

3902: %%Page: 5 5

3903: %%BeginPageSetup

3904: initializepage

3905: (rens; page: 5 of 8)setjob

3906: %%EndPageSetup

3907: -29 -30 :T

3908: gS 29 30 538 781 rC

3909: 70 96 :M

3910: f3_11 sf

3911: 3.39 .339(random derivations from the )J

3912: 222 96 :M

3913: 4.076 .408(chart and by)J

3914: 70 110 :M

3915: 1.222 .122(selecting the analysis which results most often)J

3916: 70 124 :M

3917: .217 .022(from these derivations. )J

3918: 176 124 :M

3919: .608 .061(This technique is known)J

3920: 70 138 :M

3921: .829 .083(as "Monte Carlo disambiguation" )J

3922: 227 138 :M

3923: 1.352 .135(and has been)J

3924: 70 152 :M

3925: .898 .09(extensively described in the literature \(e.g. Bod)J

3926: 70 166 :M

3927: 4.619 .462(1993, 1998; Chappelier & Rajman )J

3928: 263 166 :M

3929: -.264(2000;)A

3930: 70 180 :M

3931: 1.676 .168(Goodman 1998; Hoogweg )J

3932: 199 180 :M

3933: 1.956 .196(2000\). Sampling a)J

3934: 70 194 :M

3935: 2.105 .21(random derivation from the chart )J

3936: 237 194 :M

3937: 2.13 .213(consists of)J

3938: 70 208 :M

3939: 1.097 .11(choosing at random one of the )J

3940: 217 208 :M

3941: 1.26 .126(fragments from)J

3942: 70 222 :M

3943: .233 .023(the set of )J

3944: 115 222 :M

3945: f4_11 sf

3946: -.06(composable)A

3947: 168 222 :M

3948: f3_11 sf

3949: .281 .028( fragments at every )J

3950: 258 222 :M

3951: -.293(labeled)A

3952: 70 236 :M

3953: 1.318 .132(chart-entry \(where the random choices )J

3954: 255 236 :M

3955: 1.353 .135(at each)J

3956: 70 250 :M

3957: .782 .078(chart-entry are based on the )J

3958: 202 250 :M

3959: .748 .075(probabilities of the)J

3960: 70 264 :M

3961: -.066(fragments\). The )A

3962: 143 264 :M

3963: .139 .014(derivations are sampled in a top-)J

3964: 70 278 :M

3965: .349 .035(down, leftmost order so as to maintain the LFG)J

3966: 284 278 :M

3967: (-)S

3968: 70 292 :M

3969: .328 .033(DOP derivation order. Thus the competition sets)J

3970: 70 306 :M

3971: -.008(of composable fragments are computed on the fly)A

3972: 70 320 :M

3973: 2.21 .221(during the Monte Carlo sampling process )J

3974: 278 320 :M

3975: (by)S

3976: 70 334 :M

3977: .148 .015(grouping the f-structure )J

3978: 179 334 :M

3979: .706 .071(units that unify and that)J

3980: 70 348 :M

3981: -.11(are coherent with the subderivation built so far.)A

3982: 105 362 :M

3983: 5.974 .597(As mentioned in section )J

3984: 253 362 :M

3985: 5.708 .571(3, the)J

3986: 70 376 :M

3987: -.099(Completeness condition can only )A

3988: 217 376 :M

3989: -.07(be checked after)A

3990: 70 390 :M

3991: -.396(the )A

3992: 86 390 :M

3993: -.024(derivation process. Incomplete derivations are)A

3994: 70 404 :M

3995: .653 .065(simply removed from the sampling distribution.)J

3996: 70 418 :M

3997: 1.595 .16(After sampling a )J

3998: 155 418 :M

3999: 1.741 .174(sufficiently large number of)J

4000: 70 432 :M

4001: .468 .047(random derivations that satisfy the LFG validity)J

4002: 70 446 :M

4003: 3.038 .304(requirements, the most probable analysis )J

4004: 281 446 :M

4005: -.334(is)A

4006: 70 460 :M

4007: -.058(estimated by the analysis which results most often)A

4008: 70 474 :M

4009: -.028(from the sampled derivations. As a stop condition)A

4010: 70 488 :M

4011: 3.542 .354(on )J

4012: 89 488 :M

4013: 3.25 .325(the number of sampled derivations, we)J

4014: 70 502 :M

4015: 1.39 .139(compute the probability of error, which )J

4016: 261 502 :M

4017: 1.651 .165(is the)J

4018: 70 516 :M

4019: -.102(probability that the analysis that is most frequently)A

4020: 70 530 :M

4021: .07 .007(generated by the sampled derivations is not equal)J

4022: 70 544 :M

4023: .168 .017(to the most probable analysis, and which is set to)J

4024: 70 558 :M

4025: 1.868 .187(0.05 \(see Bod 1998\). In order to rule out )J

4026: 276 558 :M

4027: -.719(the)A

4028: 70 572 :M

4029: .194 .019(possibility that the sampling process never )J

4030: 263 572 :M

4031: -.072(stops,)A

4032: 70 586 :M

4033: 2.855 .286(we )J

4034: 90 586 :M

4035: 3.158 .316(use a maximum sample size of 10,000)J

4036: 70 600 :M

4037: -.147(derivations.)A

4038: 105 614 :M

4039: 1.443 .144(While the )J

4040: 156 614 :M

4041: 1.227 .123(Monte Carlo disambiguation)J

4042: 70 628 :M

4043: 2.86 .286(technique converges )J

4044: 175 628 :M

4045: 4.544 .454(provably to the most)J

4046: 70 642 :M

4047: 2.018 .202(probable analysis, it is quite )J

4048: 213 642 :M

4049: 1.758 .176(inefficient. It is)J

4050: 70 656 :M

4051: 1.578 .158(possible to )J

4052: 126 656 :M

4053: 1.445 .144(use an alternative, heuristic search)J

4054: 70 670 :M

4055: .209 .021(based on Viterbi )J

4056: f4_11 sf

4057: (n)S

4058: 153 670 :M

4059: f3_11 sf

4060: .313 .031( best )J

4061: 178 670 :M

4062: .966 .097(\(we will not go into the)J

4063: 70 684 :M

4064: -.058(PCFG-reduction technique presented in Goodman)A

4065: 70 698 :M

4066: .781 .078(\(1998\) since that heuristic only works for Tree)J

4067: 285 698 :M

4068: (-)S

4069: 70 712 :M

4070: 1.642 .164(DOP and is beneficial only if all )J

4071: 233 712 :M

4072: 1.327 .133(subtrees are)J

4073: 70 726 :M

4074: .544 .054(taken into account )J

4075: 157 726 :M

4076: 1.179 .118(and if the so-called "labeled)J

4077: 70 740 :M

4078: -.085(recall parse" is computed\). A Viterbi )A

4079: f4_11 sf

4080: (n)S

4081: 238 740 :M

4082: f3_11 sf

4083: -.243( best )A

4084: 261 740 :M

4085: -.217(search)A

4086: 305 96 :M

4087: 3.597 .36(for LFG-DOP estimates the )J

4088: 454 96 :M

4089: 2.445 .244(most probable)J

4090: 305 110 :M

4091: 4.954 .495(analysis by computing )J

4092: 433 110 :M

4093: f4_11 sf

4094: 1.407(n)A

4095: f3_11 sf

4096: 5.067 .507( most probable)J

4097: 305 124 :M

4098: 2.777 .278(derivations, and )J

4099: 389 124 :M

4100: 4.537 .454(by then summing up the)J

4101: 305 138 :M

4102: -.005(probabilities of the valid derivations )A

4103: 468 138 :M

4104: .157 .016(that produce)J

4105: 305 152 :M

4106: -.089(the same analysis. The algorithm )A

4107: 451 152 :M

4108: .306 .031(for computing )J

4109: f4_11 sf

4110: (n)S

4111: 305 166 :M

4112: f3_11 sf

4113: 4.9 .49(most )J

4114: 337 166 :M

4115: 3.093 .309(probable derivations follows straight)J

4116: 520 166 :M

4117: (-)S

4118: 305 180 :M

4119: -.034(forwardly from the algorithm which computes )A

4120: 511 180 :M

4121: -.719(the)A

4122: 305 194 :M

4123: 1.423 .142(most probable derivation by )J

4124: 442 194 :M

4125: 1.739 .174(means of Viterbi)J

4126: 305 208 :M

4127: -.071(optimization \(see e.g. Sima'an 1999\).)A

4128: 305 232 :M

4129: f0_12 sf

4130: -.039(5  Experimental Evaluation)A

4131: 305 246 :M

4132: f3_11 sf

4133: 2.516 .252(We derived some experimental properties )J

4134: 515 246 :M

4135: -.16(of)A

4136: 305 260 :M

4137: 1.19 .119(LFG-DOP by studying its behavior on )J

4138: 489 260 :M

4139: .821 .082(the two)J

4140: 305 274 :M

4141: 4.727 .473(LFG-annotated corpora that )J

4142: 458 274 :M

4143: 3.694 .369(are currently)J

4144: 305 288 :M

4145: 4.632 .463(available: the Verbmobil corpus and the)J

4146: 305 302 :M

4147: -.073(Homecentre corpus. )A

4148: 396 302 :M

4149: -.014(Both corpora were annotated)A

4150: 305 316 :M

4151: 2.443 .244(at Xerox PARC. They contain packed LFG)J

4152: 520 316 :M

4153: (-)S

4154: 305 330 :M

4155: .297 .03(representations \(Maxwell & Kaplan 1991\) of )J

4156: 510 330 :M

4157: -.719(the)A

4158: 305 344 :M

4159: -.096(grammatical parses of each sentence )A

4160: 466 344 :M

4161: -.112(together with)A

4162: 305 358 :M

4163: .271 .027(an indication which of these parses is )J

4164: 476 358 :M

4165: -.071(the correct)A

4166: 305 372 :M

4167: -.059(one. For our experiments we only used the correct)A

4168: 305 386 :M

4169: 3.605 .361(parses )J

4170: 342 386 :M

4171: 4.333 .433(of each sentence resulting in 540)J

4172: 305 400 :M

4173: 1.387 .139(Verbmobil parses and 980 Homecentre parses.)J

4174: 305 414 :M

4175: .403 .04(Each corpus was divided into a 90% )J

4176: 474 414 :M

4177: -.016(training set)A

4178: 305 428 :M

4179: 1.237 .124(and a 10% )J

4180: 360 428 :M

4181: 1.603 .16(test set. This division was random)J

4182: 305 442 :M

4183: -.008(except for one constraint: that all the words in the)A

4184: 305 456 :M

4185: (test set )S

4186: 339 456 :M

4187: .588 .059(actually occurred in the training set. The)J

4188: 305 470 :M

4189: 2.186 .219(sentences from the test )J

4190: 423 470 :M

4191: 2.829 .283(set were parsed and)J

4192: 305 484 :M

4193: .903 .09(disambiguated by means of the fragments from)J

4194: 305 498 :M

4195: -.063(the )A

4196: 322 498 :M

4197: .898 .09(training set. Due to memory limitations, we)J

4198: 305 512 :M

4199: 2.263 .226(restricted the maximum depth of the )J

4200: 490 512 :M

4201: -.303(indexed)A

4202: 305 526 :M

4203: 1.058 .106(subtrees to 4. Because of the )J

4204: 444 526 :M

4205: 1.353 .135(small size of the)J

4206: 305 540 :M

4207: -.048(corpora )A

4208: 342 540 :M

4209: .894 .089(we averaged our results on 10 different)J

4210: 305 554 :M

4211: .912 .091(training/test set splits. )J

4212: 410 554 :M

4213: 1.968 .197(Besides an )J

4214: 468 554 :M

4215: f4_11 sf

4216: 1.083 .108(exact match)J

4217: 305 568 :M

4218: f3_11 sf

4219: -.097(accuracy metric, we also used a )A

4220: 445 568 :M

4221: -.068(more fine-grained)A

4222: 305 582 :M

4223: 2.387 .239(score based on )J

4224: 384 582 :M

4225: 2.308 .231(the well-known PARSEVAL)J

4226: 305 596 :M

4227: -.16(metrics )A

4228: 340 596 :M

4229: -.049(that evaluate phrase-structure trees \(Black)A

4230: 305 610 :M

4231: .236 .024(et al. 1991\). The PARSEVAL metrics compare a)J

4232: 305 624 :M

4233: .287 .029(proposed parse )J

4234: 376 624 :M

4235: f4_11 sf

4236: .286(P)A

4237: f3_11 sf

4238: .45 .045( with the )J

4239: 427 624 :M

4240: -.01(corresponding correct)A

4241: 305 638 :M

4242: -.103(treebank parse )A

4243: f4_11 sf

4244: (T)S

4245: 377 638 :M

4246: f3_11 sf

4247: -.064( as follows:)A

4248: 306 662 49 10 rC

4249: gS

4250: .88 .8 scale

4251: 376.025 833.263 :T

4252: -376.025 -833.263 :T

4253: 349.025 836.263 :M

4254: f3_12 sf

4255: (Precision = )S

4256: gR

4257: gR

4258: gS 452 654 3 4 rC

4259: gS

4260: .88 .8 scale

4261: 515.011 819.263 :T

4262: -515.011 -819.263 :T

4263: 515.011 821.263 :M

4264: f4_4 sf

4265: ( )S

4266: gR

4267: gR

4268: gS 360 651 112 12 rC

4269: gS

4270: .88 .8 scale

4271: 472.417 820.013 :T

4272: -472.417 -820.013 :T

4273: 410.417 825.013 :M

4274: f3_12 sf

4275: (# correct constituents in )S

4276: gR

4277: gS

4278: .88 .8 scale

4279: 472.417 820.013 :T

4280: -472.417 -820.013 :T

4281: 528.654 825.013 :M

4282: f4_12 sf

4283: (P)S

4284: gR

4285: gR

4286: gS 375 669 80 12 rC

4287: gS

4288: .88 .8 scale

4289: 470.47 843.763 :T

4290: -470.47 -843.763 :T

4291: 427.47 848.763 :M

4292: f3_12 sf

4293: (# constituents in )S

4294: gR

4295: gS

4296: .88 .8 scale

4297: 470.47 843.763 :T

4298: -470.47 -843.763 :T

4299: 509.327 848.763 :M

4300: f4_12 sf

4301: (P  )S

4302: gR

4303: gR

4304: gS 305 650 168 32 rC

4305: 361 667 -1 1 471 666 1 361 666 @a

4306: gR

4307: gS 438 695 3 4 rC

4308: gS

4309: .881 .8 scale

4310: 498.498 870.513 :T

4311: -498.498 -870.513 :T

4312: 498.498 872.513 :M

4313: f4_4 sf

4314: ( )S

4315: gR

4316: gR

4317: gS 347 692 112 12 rC

4318: gS

4319: .881 .8 scale

4320: 457.165 871.263 :T

4321: -457.165 -871.263 :T

4322: 395.165 876.263 :M

4323: f3_12 sf

4324: (# correct constituents in )S

4325: gR

4326: gS

4327: .881 .8 scale

4328: 457.165 871.263 :T

4329: -457.165 -871.263 :T

4330: 513.26 876.263 :M

4331: f4_12 sf

4332: (P)S

4333: gR

4334: gR

4335: gS 362 710 80 12 rC

4336: gS

4337: .881 .8 scale

4338: 456.198 895.014 :T

4339: -456.198 -895.014 :T

4340: 412.198 900.014 :M

4341: f3_12 sf

4342: (# constituents in )S

4343: gR

4344: gS

4345: .881 .8 scale

4346: 456.198 895.014 :T

4347: -456.198 -895.014 :T

4348: 493.956 900.014 :M

4349: f4_12 sf

4350: (T )S

4351: gR

4352: gR

4353: gS 305 691 155 32 rC

4354: 348 708 -1 1 458 707 1 348 707 @a

4355: 306 701 36 12 rC

4356: gS

4357: .881 .8 scale

4358: 367.608 882.514 :T

4359: -367.608 -882.514 :T

4360: 348.608 887.514 :M

4361: f3_12 sf

4362: (Recall =)S

4363: gR

4364: gR

4365: gS 340 701 5 12 rC

4366: gS

4367: .881 .8 scale

4368: 388.216 882.514 :T

4369: -388.216 -882.514 :T

4370: 387.216 887.514 :M

4371: f3_12 sf

4372: ( )S

4373: gR

4374: 341 708 3 4 rC

4375: gS

4376: .881 .8 scale

4377: 388.351 886.764 :T

4378: -388.351 -886.764 :T

4379: 388.351 888.764 :M

4380: f3_4 sf

4381: ( )S

4382: gR

4383: endp

4384: %%Page: 6 6

4385: %%BeginPageSetup

4386: initializepage

4387: (rens; page: 6 of 8)setjob

4388: %%EndPageSetup

4389: -29 -30 :T

4390: gS 29 30 538 781 rC

4391: 70 96 :M

4392: f3_11 sf

4393: 1.301 .13(A constituent in )J

4394: 151 96 :M

4395: f4_11 sf

4396: .914(P)A

4397: f3_11 sf

4398: 1.747 .175( is correct if there exists a)J

4399: 70 110 :M

4400: -.023(constituent in )A

4401: 133 110 :M

4402: f4_11 sf

4403: .36(T)A

4404: f3_11 sf

4405: .376 .038( of )J

4406: 156 110 :M

4407: .829 .083(the same label that spans the)J

4408: 70 125 :M

4409: .533 .053(same words and that )J

4410: 167 125 :M

4411: f1_11 sf

4412: .151(f)A

4413: f3_11 sf

4414: .712 .071(-corresponds )J

4415: 235 125 :M

4416: .969 .097(to the same)J

4417: 70 140 :M

4418: 2.121 .212(f-structure )J

4419: 124 140 :M

4420: 4.439 .444(unit \(see Bod 2000c for some)J

4421: 70 154 :M

4422: -.034(illustrations of these metrics for LFG-DOP\).)A

4423: 70 177 :M

4424: f0_11 sf

4425: -.006(5.1 Comparing the two fragment estimators)A

4426: 70 191 :M

4427: f3_11 sf

4428: 3.816 .382(We were first )J

4429: 150 191 :M

4430: 3.029 .303(interested in comparing the)J

4431: 70 205 :M

4432: -.029(performance )A

4433: 129 205 :M

4434: .818 .082(of the simple RF estimator against)J

4435: 70 219 :M

4436: 1.32 .132(the discounted RF )J

4437: 161 219 :M

4438: 1.394 .139(estimator. Furthermore, we)J

4439: 70 233 :M

4440: 2.42 .242(want to study the )J

4441: 163 233 :M

4442: 1.062 .106(contribution of generalized)J

4443: 70 247 :M

4444: 1.373 .137(fragments to the parse accuracy. )J

4445: 227 247 :M

4446: 1.096 .11(We therefore)J

4447: 70 261 :M

4448: -.105(created for each training set two sets of )A

4449: 242 261 :M

4450: -.115(fragments:)A

4451: 70 275 :M

4452: -.037(one which contains )A

4453: 158 275 :M

4454: f4_11 sf

4455: .087(all)A

4456: f3_11 sf

4457: .345 .034( fragments \(up to depth 4\))J

4458: 70 289 :M

4459: -.079(and one which excludes the generalized fragments)A

4460: 70 303 :M

4461: .187 .019(as generated by )J

4462: 143 303 :M

4463: f4_11 sf

4464: -.073(Discard)A

4465: 179 303 :M

4466: f3_11 sf

4467: .513 .051(. The exclusion of these)J

4468: 70 317 :M

4469: f4_11 sf

4470: -.073(Discard)A

4471: 106 317 :M

4472: f3_11 sf

4473: 3.525 .352(-generated fragments means that )J

4474: 278 317 :M

4475: -.497(all)A

4476: 70 331 :M

4477: .419 .042(probability mass goes to the fragments )J

4478: 248 331 :M

4479: -.218(generated)A

4480: 70 345 :M

4481: 1.875 .187(by )J

4482: 87 345 :M

4483: f4_11 sf

4484: .597(Root)A

4485: f3_11 sf

4486: 1.023 .102( and )J

4487: 135 345 :M

4488: f4_11 sf

4489: -.039(Frontier)A

4490: 173 345 :M

4491: f3_11 sf

4492: 2.247 .225( in which case the two)J

4493: 70 359 :M

4494: 2.242 .224(estimators are equivalent. The following )J

4495: 272 359 :M

4496: -.249(two)A

4497: 70 373 :M

4498: 1.496 .15(tables present )J

4499: 140 373 :M

4500: 2.504 .25(the results of our experiments)J

4501: 70 387 :M

4502: .084 .008(where +Discard refers to the full set of fragments)J

4503: 70 402 :M

4504: .178 .018(and )J

4505: f1_11 sf

4506: (-)S

4507: 96 402 :M

4508: f3_11 sf

4509: .867 .087(Discard refers to the fragment set without)J

4510: 70 417 :M

4511: f4_11 sf

4512: -.073(Discard)A

4513: 106 417 :M

4514: f3_11 sf

4515: -.094(-generated fragments.)A

4516: 72 436 14 11 rC

4517: gS

4518: .537 .745 scale

4519: 145.998 591.965 :T

4520: -145.998 -591.965 :T

4521: 133.998 596.965 :M

4522: f3_12 sf

4523: (        )S

4524: gR

4525: gR

4526: gS 128 436 44 11 rC

4527: gS

4528: .537 .745 scale

4529: 279.079 591.965 :T

4530: -279.079 -591.965 :T

4531: 240.079 596.965 :M

4532: f3_12 sf

4533: (      Exact Match)S

4534: gR

4535: gR

4536: gS 184 436 43 11 rC

4537: gS

4538: .537 .745 scale

4539: 383.299 591.965 :T

4540: -383.299 -591.965 :T

4541: 344.299 596.965 :M

4542: f3_12 sf

4543: (        Precision   )S

4544: gR

4545: gR

4546: gS 260 436 17 11 rC

4547: gS

4548: .537 .745 scale

4549: 498.88 591.965 :T

4550: -498.88 -591.965 :T

4551: 483.88 596.965 :M

4552: f3_12 sf

4553: (Recall)S

4554: gR

4555: gR

4556: gS 128 448 28 11 rC

4557: gS

4558: .537 .745 scale

4559: 264.079 608.063 :T

4560: -264.079 -608.063 :T

4561: 240.079 613.063 :M

4562: f3_12 sf

4563: (+Discard  )S

4564: gR

4565: gR

4566: gS 155 447 5 13 rC

4567: gS

4568: .537 .745 scale

4569: 291.467 608.063 :T

4570: -291.467 -608.063 :T

4571: 288.467 613.063 :M

4572: f1_12 sf

4573: (-)S

4574: gR

4575: gR

4576: gS 159 448 21 11 rC

4577: gS

4578: .537 .745 scale

4579: 313.911 608.063 :T

4580: -313.911 -608.063 :T

4581: 295.911 613.063 :M

4582: f3_12 sf

4583: (Discard)S

4584: gR

4585: gR

4586: gS 184 448 28 11 rC

4587: gS

4588: .537 .745 scale

4589: 369.299 608.063 :T

4590: -369.299 -608.063 :T

4591: 344.299 613.063 :M

4592: f3_12 sf

4593: (+Discard  )S

4594: gR

4595: gR

4596: gS 211 447 5 13 rC

4597: gS

4598: .537 .745 scale

4599: 397.549 608.063 :T

4600: -397.549 -608.063 :T

4601: 394.549 613.063 :M

4602: f1_12 sf

4603: (-)S

4604: gR

4605: gR

4606: gS 215 448 21 11 rC

4607: gS

4608: .537 .745 scale

4609: 419.993 608.063 :T

4610: -419.993 -608.063 :T

4611: 401.993 613.063 :M

4612: f3_12 sf

4613: (Discard)S

4614: gR

4615: gR

4616: gS 241 448 28 11 rC

4617: gS

4618: .537 .745 scale

4619: 472.52 608.063 :T

4620: -472.52 -608.063 :T

4621: 448.52 613.063 :M

4622: f3_12 sf

4623: (+Discard  )S

4624: gR

4625: gR

4626: gS 268 447 5 13 rC

4627: gS

4628: .537 .745 scale

4629: 501.769 608.063 :T

4630: -501.769 -608.063 :T

4631: 498.769 613.063 :M

4632: f1_12 sf

4633: (-)S

4634: gR

4635: gR

4636: gS 272 448 20 11 rC

4637: gS

4638: .537 .745 scale

4639: 524.213 608.063 :T

4640: -524.213 -608.063 :T

4641: 506.213 613.063 :M

4642: f3_12 sf

4643: (Discard)S

4644: gR

4645: gR

4646: gS 73 471 41 13 rC

4647: gS

4648: .537 .745 scale

4649: 172.859 639.917 :T

4650: -172.859 -639.917 :T

4651: 135.859 643.917 :M

4652: f3_12 sf

4653: (Simple RF)S

4654: gR

4655: gR

4656: gS 130 471 15 11 rC

4657: gS

4658: .537 .745 scale

4659: 255.801 638.917 :T

4660: -255.801 -638.917 :T

4661: 243.801 643.917 :M

4662: f3_12 sf

4663: (1.1%)S

4664: gR

4665: gR

4666: gS 150 471 27 11 rC

4667: gS

4668: .537 .745 scale

4669: 305.023 638.917 :T

4670: -305.023 -638.917 :T

4671: 281.023 643.917 :M

4672: f3_12 sf

4673: (      35.2%)S

4674: gR

4675: gR

4676: gS 188 471 47 11 rC

4677: gS

4678: .537 .745 scale

4679: 392.883 638.917 :T

4680: -392.883 -638.917 :T

4681: 349.883 643.917 :M

4682: f3_12 sf

4683: (13.8%        76.0%)S

4684: gR

4685: gR

4686: gS 246 471 45 11 rC

4687: gS

4688: .537 .745 scale

4689: 498.825 638.917 :T

4690: -498.825 -638.917 :T

4691: 457.825 643.917 :M

4692: f3_12 sf

4693: (11.5%       74.9%)S

4694: gR

4695: gR

4696: gS 130 493 47 11 rC

4697: gS

4698: .537 .745 scale

4699: 285.801 668.43 :T

4700: -285.801 -668.43 :T

4701: 243.801 673.43 :M

4702: f3_12 sf

4703: (35.9%        35.2%)S

4704: gR

4705: gR

4706: gS 188 493 47 11 rC

4707: gS

4708: .537 .745 scale

4709: 392.883 668.43 :T

4710: -392.883 -668.43 :T

4711: 349.883 673.43 :M

4712: f3_12 sf

4713: (77.5%        76.0%)S

4714: gR

4715: gR

4716: gS 246 492 45 12 rC

4717: gS

4718: .537 .745 scale

4719: 498.825 667.088 :T

4720: -498.825 -667.088 :T

4721: 457.825 672.088 :M

4722: f3_12 sf

4723: (76.4%       74.9%)S

4724: gR

4725: gR

4726: gS 70 429 223 82 rC

4727: 71 431 -1 1 291 430 1 71 430 @a

4728: 72 464 -1 1 292 463 1 72 463 @a

4729: 72 511 -1 1 292 510 1 72 510 @a

4730: 72 493 40 14 rC

4731: gS

4732: .537 .745 scale

4733: 170.859 669.43 :T

4734: -170.859 -669.43 :T

4735: 135.859 673.43 :M

4736: f3_12 sf

4737: (Discounted RF)S

4738: gR

4739: gR

4740: gS 72 444 25 9 rC

4741: gS

4742: .537 .745 scale

4743: 156.998 601.014 :T

4744: -156.998 -601.014 :T

4745: 133.998 605.014 :M

4746: f3_12 sf

4747: (Estimator)S

4748: gR

4749: gR

4750: gS 29 30 538 781 rC

4751: 75 531 :M

4752: f3_11 sf

4753: -.063(Table 1. Experimental results on the Verbmobil)A

4754: 72 554 14 11 rC

4755: gS

4756: .537 .718 scale

4757: 145.998 778.953 :T

4758: -145.998 -778.953 :T

4759: 133.998 783.953 :M

4760: f3_12 sf

4761: (        )S

4762: gR

4763: gR

4764: gS 128 554 44 11 rC

4765: gS

4766: .537 .718 scale

4767: 279.079 778.953 :T

4768: -279.079 -778.953 :T

4769: 240.079 783.953 :M

4770: f3_12 sf

4771: (      Exact Match)S

4772: gR

4773: gR

4774: gS 184 554 43 11 rC

4775: gS

4776: .537 .718 scale

4777: 383.299 778.953 :T

4778: -383.299 -778.953 :T

4779: 344.299 783.953 :M

4780: f3_12 sf

4781: (        Precision   )S

4782: gR

4783: gR

4784: gS 260 554 17 11 rC

4785: gS

4786: .537 .718 scale

4787: 498.88 778.953 :T

4788: -498.88 -778.953 :T

4789: 483.88 783.953 :M

4790: f3_12 sf

4791: (Recall)S

4792: gR

4793: gR

4794: gS 128 566 28 11 rC

4795: gS

4796: .537 .718 scale

4797: 264.079 795.663 :T

4798: -264.079 -795.663 :T

4799: 240.079 800.663 :M

4800: f3_12 sf

4801: (+Discard  )S

4802: gR

4803: gR

4804: gS 155 565 5 12 rC

4805: gS

4806: .537 .718 scale

4807: 291.467 795.663 :T

4808: -291.467 -795.663 :T

4809: 288.467 800.663 :M

4810: f1_12 sf

4811: (-)S

4812: gR

4813: gR

4814: gS 159 566 21 11 rC

4815: gS

4816: .537 .718 scale

4817: 313.911 795.663 :T

4818: -313.911 -795.663 :T

4819: 295.911 800.663 :M

4820: f3_12 sf

4821: (Discard)S

4822: gR

4823: gR

4824: gS 184 566 28 11 rC

4825: gS

4826: .537 .718 scale

4827: 369.299 795.663 :T

4828: -369.299 -795.663 :T

4829: 344.299 800.663 :M

4830: f3_12 sf

4831: (+Discard  )S

4832: gR

4833: gR

4834: gS 211 565 5 12 rC

4835: gS

4836: .537 .718 scale

4837: 397.549 795.663 :T

4838: -397.549 -795.663 :T

4839: 394.549 800.663 :M

4840: f1_12 sf

4841: (-)S

4842: gR

4843: gR

4844: gS 215 566 21 11 rC

4845: gS

4846: .537 .718 scale

4847: 419.993 795.663 :T

4848: -419.993 -795.663 :T

4849: 401.993 800.663 :M

4850: f3_12 sf

4851: (Discard)S

4852: gR

4853: gR

4854: gS 241 566 28 11 rC

4855: gS

4856: .537 .718 scale

4857: 472.52 795.663 :T

4858: -472.52 -795.663 :T

4859: 448.52 800.663 :M

4860: f3_12 sf

4861: (+Discard  )S

4862: gR

4863: gR

4864: gS 268 565 5 12 rC

4865: gS

4866: .537 .718 scale

4867: 501.769 795.663 :T

4868: -501.769 -795.663 :T

4869: 498.769 800.663 :M

4870: f1_12 sf

4871: (-)S

4872: gR

4873: gR

4874: gS 272 566 20 11 rC

4875: gS

4876: .537 .718 scale

4877: 524.213 795.663 :T

4878: -524.213 -795.663 :T

4879: 506.213 800.663 :M

4880: f3_12 sf

4881: (Discard)S

4882: gR

4883: gR

4884: gS 131 588 14 12 rC

4885: gS

4886: .537 .718 scale

4887: 255.801 826.297 :T

4888: -255.801 -826.297 :T

4889: 243.801 831.297 :M

4890: f3_12 sf

4891: (2.7%)S

4892: gR

4893: gR

4894: gS 150 588 27 12 rC

4895: gS

4896: .537 .718 scale

4897: 305.023 826.297 :T

4898: -305.023 -826.297 :T

4899: 281.023 831.297 :M

4900: f3_12 sf

4901: (      37.9%)S

4902: gR

4903: gR

4904: gS 188 588 47 12 rC

4905: gS

4906: .537 .718 scale

4907: 392.883 826.297 :T

4908: -392.883 -826.297 :T

4909: 349.883 831.297 :M

4910: f3_12 sf

4911: (17.1%        77.8%)S

4912: gR

4913: gR

4914: gS 246 588 45 12 rC

4915: gS

4916: .537 .718 scale

4917: 498.825 826.297 :T

4918: -498.825 -826.297 :T

4919: 457.825 831.297 :M

4920: f3_12 sf

4921: (15.5%       77.2%)S

4922: gR

4923: gR

4924: gS 130 609 47 12 rC

4925: gS

4926: .537 .718 scale

4927: 285.801 856.539 :T

4928: -285.801 -856.539 :T

4929: 243.801 860.539 :M

4930: f3_12 sf

4931: (38.4%        37.9%)S

4932: gR

4933: gR

4934: gS 188 609 47 12 rC

4935: gS

4936: .537 .718 scale

4937: 392.883 856.539 :T

4938: -392.883 -856.539 :T

4939: 349.883 860.539 :M

4940: f3_12 sf

4941: (80.0%        77.8%)S

4942: gR

4943: gR

4944: gS 246 609 45 12 rC

4945: gS

4946: .537 .718 scale

4947: 498.825 856.539 :T

4948: -498.825 -856.539 :T

4949: 457.825 860.539 :M

4950: f3_12 sf

4951: (78.6%       77.2%)S

4952: gR

4953: gR

4954: gS 70 548 223 79 rC

4955: 71 550 -1 1 291 549 1 71 549 @a

4956: 72 582 -1 1 292 581 1 72 581 @a

4957: 72 627 -1 1 292 626 1 72 626 @a

4958: 72 589 42 11 rC

4959: gS

4960: .537 .718 scale

4961: 172.859 828.689 :T

4962: -172.859 -828.689 :T

4963: 135.859 832.689 :M

4964: f3_12 sf

4965: (Simple RF)S

4966: gR

4967: gR

4968: gS 72 609 40 12 rC

4969: gS

4970: .537 .718 scale

4971: 170.859 856.539 :T

4972: -170.859 -856.539 :T

4973: 135.859 860.539 :M

4974: f3_12 sf

4975: (Discounted RF)S

4976: gR

4977: gR

4978: gS 71 562 26 8 rC

4979: gS

4980: .537 .718 scale

4981: 155.998 787.916 :T

4982: -155.998 -787.916 :T

4983: 133.998 790.916 :M

4984: f3_12 sf

4985: (Estimator)S

4986: gR

4987: gR

4988: gS 29 30 538 781 rC

4989: 73 647 :M

4990: f3_11 sf

4991: -.086(Table 2. Experimental results on the Homecentre)A

4992: 70 670 :M

4993: 1.567 .157(The tables show that the simple )J

4994: 228 670 :M

4995: 1.367 .137(RF estimator)J

4996: 70 684 :M

4997: -.062(scores extremely bad if all fragments are used: the)A

4998: 70 698 :M

4999: 2.485 .248(exact match is only 1.1% on )J

5000: 220 698 :M

5001: 1.436 .144(the Verbmobil)J

5002: 70 712 :M

5003: 2.458 .246(corpus and 2.7% on the )J

5004: 196 712 :M

5005: 1.273 .127(Homecentre corpus,)J

5006: 70 726 :M

5007: 2.592 .259(whereas the )J

5008: 134 726 :M

5009: 2.752 .275(discounted RF estimator scores)J

5010: 70 740 :M

5011: .062 .006(respectively 35.9% and )J

5012: 177 740 :M

5013: .847 .085(38.4% on these corpora.)J

5014: 305 96 :M

5015: 1.148 .115(Also the more )J

5016: 376 96 :M

5017: .547 .055(fine-grained precision and recall)J

5018: 305 110 :M

5019: .194 .019(scores obtained with the simple RF estimator are)J

5020: 305 124 :M

5021: 3.317 .332(quite low: )J

5022: 363 124 :M

5023: 4.372 .437(e.g. 13.8% and 11.5% on the)J

5024: 305 138 :M

5025: 2.565 .256(Verbmobil corpus, where the discounted )J

5026: 510 138 :M

5027: -.452(RF)A

5028: 305 152 :M

5029: -.074(estimator obtains 77.5% and )A

5030: 432 152 :M

5031: -.062(76.4%. Interestingly,)A

5032: 305 166 :M

5033: .315 .032(the accuracy of the simple RF estimator is )J

5034: 498 166 :M

5035: -.146(much)A

5036: 305 180 :M

5037: 5.408 .541(higher if )J

5038: 362 180 :M

5039: f4_11 sf

5040: -.073(Discard)A

5041: 398 180 :M

5042: f3_11 sf

5043: 3.779 .378(-generated fragments )J

5044: 511 180 :M

5045: -.713(are)A

5046: 305 194 :M

5047: .561 .056(excluded. This suggests that treating )J

5048: 475 194 :M

5049: -.268(generalized)A

5050: 305 208 :M

5051: .67 .067(fragments )J

5052: 354 208 :M

5053: 1.258 .126(probabilistically in the same way as)J

5054: 305 222 :M

5055: -.083(ungeneralized fragments is harmful.)A

5056: 340 236 :M

5057: -.009(The tables also show that the inclusion of)A

5058: 305 250 :M

5059: f4_11 sf

5060: -.073(Discard)A

5061: 341 250 :M

5062: f3_11 sf

5063: 2.319 .232(-generated fragments leads only )J

5064: 502 250 :M

5065: 3.701 .37(to a)J

5066: 305 264 :M

5067: .027 .003(slight accuracy increase under the )J

5068: 458 264 :M

5069: .519 .052(discounted RF)J

5070: 305 278 :M

5071: .992 .099(estimator. Unfortunately, according to paired )J

5072: f4_11 sf

5073: (t)S

5074: 520 278 :M

5075: f3_11 sf

5076: (-)S

5077: 305 292 :M

5078: 1.755 .175(testing only )J

5079: 366 292 :M

5080: 2.242 .224(the differences for the precision)J

5081: 305 306 :M

5082: 5.959 .596(scores on the Homecentre corpus )J

5083: 503 306 :M

5084: -.122(were)A

5085: 305 320 :M

5086: -.161(statistically significant.)A

5087: 305 343 :M

5088: f0_11 sf

5089: -.032(5.2 Comparing different fragment sizes)A

5090: 305 357 :M

5091: f3_11 sf

5092: -.039(We were also interested in the impact of fragment)A

5093: 305 371 :M

5094: 4.431 .443(size on the parse )J

5095: 407 371 :M

5096: 3.105 .31(accuracy. We therefore)J

5097: 305 385 :M

5098: 2.122 .212(performed a series of experiments )J

5099: 476 385 :M

5100: 1.636 .164(where the)J

5101: 305 399 :M

5102: -.013(fragment set is restricted to fragments of a certain)A

5103: 305 413 :M

5104: .475 .048(maximum depth \(where the depth )J

5105: 461 413 :M

5106: .862 .086(of a fragment)J

5107: 305 427 :M

5108: -.133(is defined as )A

5109: 362 427 :M

5110: .312 .031(the longest path from root to leaf of)J

5111: 305 441 :M

5112: 4.909 .491(its c-structure unit\). We used the same)J

5113: 305 455 :M

5114: 4.569 .457(training/test set splits as in )J

5115: 462 455 :M

5116: 3.302 .33(the previous)J

5117: 305 469 :M

5118: 2.099 .21(experiments and used both )J

5119: 440 469 :M

5120: .648 .065(ungeneralized and)J

5121: 305 483 :M

5122: 5.833 .583(generalized fragments together with )J

5123: 511 483 :M

5124: -.719(the)A

5125: 305 497 :M

5126: -.056(discounted RF estimator.)A

5127: 307 517 63 11 rC

5128: gS

5129: .781 .768 scale

5130: 433.402 679.469 :T

5131: -433.402 -679.469 :T

5132: 394.402 684.469 :M

5133: f3_12 sf

5134: (Fragment Depth)S

5135: gR

5136: gR

5137: gS 379 517 49 11 rC

5138: gS

5139: .781 .768 scale

5140: 516.6 679.469 :T

5141: -516.6 -679.469 :T

5142: 486.6 684.469 :M

5143: f3_12 sf

5144: (Exact Match)S

5145: gR

5146: gR

5147: gS 440 517 36 11 rC

5148: gS

5149: .781 .768 scale

5150: 586.712 679.469 :T

5151: -586.712 -679.469 :T

5152: 564.712 684.469 :M

5153: f3_12 sf

5154: (Precision)S

5155: gR

5156: gR

5157: gS 499 517 25 11 rC

5158: gS

5159: .781 .768 scale

5160: 654.263 679.469 :T

5161: -654.263 -679.469 :T

5162: 640.263 684.469 :M

5163: f3_12 sf

5164: (Recall)S

5165: gR

5166: gR

5167: gS 334 540 6 11 rC

5168: gS

5169: .781 .768 scale

5170: 431.976 709.399 :T

5171: -431.976 -709.399 :T

5172: 428.976 714.399 :M

5173: f3_12 sf

5174: (1)S

5175: gR

5176: gR

5177: gS 390 540 26 11 rC

5178: gS

5179: .781 .768 scale

5180: 515.685 709.399 :T

5181: -515.685 -709.399 :T

5182: 500.685 714.399 :M

5183: f3_12 sf

5184: (30.6%)S

5185: gR

5186: gR

5187: gS 445 540 26 11 rC

5188: gS

5189: .781 .768 scale

5190: 586.114 709.399 :T

5191: -586.114 -709.399 :T

5192: 571.114 714.399 :M

5193: f3_12 sf

5194: (74.2%)S

5195: gR

5196: gR

5197: gS 499 540 26 11 rC

5198: gS

5199: .781 .768 scale

5200: 655.263 709.399 :T

5201: -655.263 -709.399 :T

5202: 640.263 714.399 :M

5203: f3_12 sf

5204: (72.2%)S

5205: gR

5206: gR

5207: gS 334 551 12 12 rC

5208: gS

5209: .781 .768 scale

5210: 434.976 723.713 :T

5211: -434.976 -723.713 :T

5212: 428.976 728.713 :M

5213: f3_12 sf

5214: cF

5215: f1_12 sf

5216: <A3>S

5217: sf

5218: (2)S

5219: gR

5220: gR

5221: gS 390 551 26 12 rC

5222: gS

5223: .781 .768 scale

5224: 515.685 723.713 :T

5225: -515.685 -723.713 :T

5226: 500.685 728.713 :M

5227: f3_12 sf

5228: (34.1%)S

5229: gR

5230: gR

5231: gS 445 551 26 12 rC

5232: gS

5233: .781 .768 scale

5234: 586.114 723.713 :T

5235: -586.114 -723.713 :T

5236: 571.114 728.713 :M

5237: f3_12 sf

5238: (76.2%)S

5239: gR

5240: gR

5241: gS 499 551 26 12 rC

5242: gS

5243: .781 .768 scale

5244: 655.263 723.713 :T

5245: -655.263 -723.713 :T

5246: 640.263 728.713 :M

5247: f3_12 sf

5248: (74.5%)S

5249: gR

5250: gR

5251: gS 334 563 12 11 rC

5252: gS

5253: .781 .768 scale

5254: 434.976 739.328 :T

5255: -434.976 -739.328 :T

5256: 428.976 744.328 :M

5257: f3_12 sf

5258: cF

5259: f1_12 sf

5260: <A3>S

5261: sf

5262: (3)S

5263: gR

5264: gR

5265: gS 390 563 26 11 rC

5266: gS

5267: .781 .768 scale

5268: 515.685 739.328 :T

5269: -515.685 -739.328 :T

5270: 500.685 744.328 :M

5271: f3_12 sf

5272: (35.6%)S

5273: gR

5274: gR

5275: gS 445 563 26 11 rC

5276: gS

5277: .781 .768 scale

5278: 586.114 739.328 :T

5279: -586.114 -739.328 :T

5280: 571.114 744.328 :M

5281: f3_12 sf

5282: (76.8%)S

5283: gR

5284: gR

5285: gS 499 563 26 11 rC

5286: gS

5287: .781 .768 scale

5288: 655.263 739.328 :T

5289: -655.263 -739.328 :T

5290: 640.263 744.328 :M

5291: f3_12 sf

5292: (75.9%)S

5293: gR

5294: gR

5295: gS 334 574 12 12 rC

5296: gS

5297: .781 .768 scale

5298: 434.976 754.943 :T

5299: -434.976 -754.943 :T

5300: 428.976 759.943 :M

5301: f3_12 sf

5302: cF

5303: f1_12 sf

5304: <A3>S

5305: sf

5306: (4)S

5307: gR

5308: gR

5309: gS 390 574 26 12 rC

5310: gS

5311: .781 .768 scale

5312: 515.685 754.943 :T

5313: -515.685 -754.943 :T

5314: 500.685 759.943 :M

5315: f3_12 sf

5316: (35.9%)S

5317: gR

5318: gR

5319: gS 445 574 26 12 rC

5320: gS

5321: .781 .768 scale

5322: 586.114 754.943 :T

5323: -586.114 -754.943 :T

5324: 571.114 759.943 :M

5325: f3_12 sf

5326: (77.5%)S

5327: gR

5328: gR

5329: gS 499 574 26 12 rC

5330: gS

5331: .781 .768 scale

5332: 655.263 754.943 :T

5333: -655.263 -754.943 :T

5334: 640.263 759.943 :M

5335: f3_12 sf

5336: (76.4%)S

5337: gR

5338: gR

5339: gS 305 509 221 83 rC

5340: 307 512 -1 1 526 511 1 307 511 @a

5341: 307 535 -1 1 526 534 1 307 534 @a

5342: 307 592 -1 1 526 591 1 307 591 @a

5343: gR

5344: gS 29 30 538 781 rC

5345: 331 612 :M

5346: f3_11 sf

5347: -.085(Table 3. Accuracies on the Verbmobil)A

5348: 307 637 63 11 rC

5349: gS

5350: .781 .778 scale

5351: 433.402 825.579 :T

5352: -433.402 -825.579 :T

5353: 394.402 830.579 :M

5354: f3_12 sf

5355: (Fragment Depth)S

5356: gR

5357: gR

5358: gS 379 637 49 11 rC

5359: gS

5360: .781 .778 scale

5361: 516.6 825.579 :T

5362: -516.6 -825.579 :T

5363: 486.6 830.579 :M

5364: f3_12 sf

5365: (Exact Match)S

5366: gR

5367: gR

5368: gS 440 637 36 11 rC

5369: gS

5370: .781 .778 scale

5371: 586.712 825.579 :T

5372: -586.712 -825.579 :T

5373: 564.712 830.579 :M

5374: f3_12 sf

5375: (Precision)S

5376: gR

5377: gR

5378: gS 499 637 25 11 rC

5379: gS

5380: .781 .778 scale

5381: 654.263 825.579 :T

5382: -654.263 -825.579 :T

5383: 640.263 830.579 :M

5384: f3_12 sf

5385: (Recall)S

5386: gR

5387: gR

5388: gS 305 629 221 84 rC

5389: 307 632 -1 1 526 631 1 307 631 @a

5390: 307 655 -1 1 526 654 1 307 654 @a

5391: 307 713 -1 1 526 712 1 307 712 @a

5392: 335 660 6 12 rC

5393: gS

5394: .781 .778 scale

5395: 430.976 855.15 :T

5396: -430.976 -855.15 :T

5397: 428.976 860.15 :M

5398: f3_12 sf

5399: (1)S

5400: gR

5401: gR

5402: gS 389 660 26 12 rC

5403: gS

5404: .781 .778 scale

5405: 514.405 855.15 :T

5406: -514.405 -855.15 :T

5407: 499.405 860.15 :M

5408: f3_12 sf

5409: (31.3%)S

5410: gR

5411: gR

5412: gS 444 660 26 12 rC

5413: gS

5414: .781 .778 scale

5415: 584.834 855.15 :T

5416: -584.834 -855.15 :T

5417: 569.834 860.15 :M

5418: f3_12 sf

5419: (75.0%)S

5420: gR

5421: gR

5422: gS 499 660 25 12 rC

5423: gS

5424: .781 .778 scale

5425: 653.982 855.15 :T

5426: -653.982 -855.15 :T

5427: 638.982 860.15 :M

5428: f3_12 sf

5429: (71.5%)S

5430: gR

5431: gR

5432: gS 335 672 11 11 rC

5433: gS

5434: .781 .778 scale

5435: 434.976 870.579 :T

5436: -434.976 -870.579 :T

5437: 428.976 875.579 :M

5438: f3_12 sf

5439: cF

5440: f1_12 sf

5441: <A3>S

5442: sf

5443: (2)S

5444: gR

5445: gR

5446: gS 389 672 26 11 rC

5447: gS

5448: .781 .778 scale

5449: 514.405 870.579 :T

5450: -514.405 -870.579 :T

5451: 499.405 875.579 :M

5452: f3_12 sf

5453: (36.3%)S

5454: gR

5455: gR

5456: gS 444 672 26 11 rC

5457: gS

5458: .781 .778 scale

5459: 584.834 870.579 :T

5460: -584.834 -870.579 :T

5461: 569.834 875.579 :M

5462: f3_12 sf

5463: (77.1%)S

5464: gR

5465: gR

5466: gS 499 672 25 11 rC

5467: gS

5468: .781 .778 scale

5469: 653.982 870.579 :T

5470: -653.982 -870.579 :T

5471: 638.982 875.579 :M

5472: f3_12 sf

5473: (74.7%)S

5474: gR

5475: gR

5476: gS 335 683 11 12 rC

5477: gS

5478: .781 .778 scale

5479: 434.976 886.008 :T

5480: -434.976 -886.008 :T

5481: 428.976 891.008 :M

5482: f3_12 sf

5483: cF

5484: f1_12 sf

5485: <A3>S

5486: sf

5487: (3)S

5488: gR

5489: gR

5490: gS 389 683 26 12 rC

5491: gS

5492: .781 .778 scale

5493: 514.405 886.008 :T

5494: -514.405 -886.008 :T

5495: 499.405 891.008 :M

5496: f3_12 sf

5497: (37.8%)S

5498: gR

5499: gR

5500: gS 444 683 26 12 rC

5501: gS

5502: .781 .778 scale

5503: 584.834 886.008 :T

5504: -584.834 -886.008 :T

5505: 569.834 891.008 :M

5506: f3_12 sf

5507: (77.8%)S

5508: gR

5509: gR

5510: gS 499 683 25 12 rC

5511: gS

5512: .781 .778 scale

5513: 653.982 886.008 :T

5514: -653.982 -886.008 :T

5515: 638.982 891.008 :M

5516: f3_12 sf

5517: (76.1%)S

5518: gR

5519: gR

5520: gS 335 695 11 12 rC

5521: gS

5522: .781 .778 scale

5523: 434.976 900.151 :T

5524: -434.976 -900.151 :T

5525: 428.976 905.151 :M

5526: f3_12 sf

5527: cF

5528: f1_12 sf

5529: <A3>S

5530: sf

5531: (4)S

5532: gR

5533: gR

5534: gS 389 695 26 12 rC

5535: gS

5536: .781 .778 scale

5537: 514.405 900.151 :T

5538: -514.405 -900.151 :T

5539: 499.405 905.151 :M

5540: f3_12 sf

5541: (38.4%)S

5542: gR

5543: gR

5544: gS 444 695 26 12 rC

5545: gS

5546: .781 .778 scale

5547: 584.834 900.151 :T

5548: -584.834 -900.151 :T

5549: 569.834 905.151 :M

5550: f3_12 sf

5551: (80.0%)S

5552: gR

5553: gR

5554: gS 499 695 25 12 rC

5555: gS

5556: .781 .778 scale

5557: 653.982 900.151 :T

5558: -653.982 -900.151 :T

5559: 638.982 905.151 :M

5560: f3_12 sf

5561: (78.6%)S

5562: gR

5563: gR

5564: gS 29 30 538 781 rC

5565: 329 733 :M

5566: f3_11 sf

5567: -.113(Table 4. Accuracies on the Homecentre)A

5568: endp

5569: %%Page: 7 7

5570: %%BeginPageSetup

5571: initializepage

5572: (rens; page: 7 of 8)setjob

5573: %%EndPageSetup

5574: -29 -30 :T

5575: gS 29 30 538 781 rC

5576: 70 96 :M

5577: f3_11 sf

5578: 1.074 .107(Tables 3 and 4 )J

5579: 144 96 :M

5580: 1.543 .154(show that there is a consistent)J

5581: 70 110 :M

5582: -.189(increase in )A

5583: 119 110 :M

5584: .069 .007(parse accuracy for all metrics if larger)J

5585: 70 124 :M

5586: -.025(fragments are included, but that the increase itself)A

5587: 70 138 :M

5588: -.06(decreases. This phenomenon is also )A

5589: 229 138 :M

5590: .31 .031(known as the)J

5591: 70 152 :M

5592: 3.114 .311(DOP hypothesis \(Bod 1998\), and )J

5593: 245 152 :M

5594: 2.838 .284(has been)J

5595: 70 166 :M

5596: .156 .016(confirmed for Tree-DOP on the ATIS, OVIS and)J

5597: 70 180 :M

5598: 1.12 .112(Wall Street )J

5599: 127 180 :M

5600: 1.928 .193(Journal treebanks \(see Bod 1993,)J

5601: 70 194 :M

5602: -.033(1998, 1999, 2000a; Sima'an 1999; Bonnema et al.)A

5603: 70 208 :M

5604: .796 .08(1997; )J

5605: 100 208 :M

5606: 1.502 .15(Hoogweg 2000\). The current result thus)J

5607: 70 222 :M

5608: 1.559 .156(extends the validity of the DOP hypothesis )J

5609: 280 222 :M

5610: -.556(to)A

5611: 70 236 :M

5612: .365 .036(LFG )J

5613: 95 236 :M

5614: .961 .096(annotations. We do not yet know whether)J

5615: 70 250 :M

5616: -.063(the )A

5617: 87 250 :M

5618: .536 .054(accuracy continues to increase if even larger)J

5619: 70 264 :M

5620: -.007(fragments are included \(for Tree-DOP it has been)A

5621: 70 278 :M

5622: .102 .01(shown that the accuracy )J

5623: f4_11 sf

5624: .033(decreases)A

5625: 224 278 :M

5626: f3_11 sf

5627: -.013( after a certain)A

5628: 70 292 :M

5629: -.119(depth, probably )A

5630: 141 292 :M

5631: .197 .02(due to overfitting -- cf. Bonnema)J

5632: 70 306 :M

5633: -.101(et al. 1997; Bod 2000a\).)A

5634: 70 329 :M

5635: f0_11 sf

5636: .551 .055(5.3 Comparing LFG-DOP to Tree-DOP)J

5637: 70 343 :M

5638: f3_11 sf

5639: .28 .028(In the following experiment, we are )J

5640: 234 343 :M

5641: -.005(interested in)A

5642: 70 357 :M

5643: -.063(the )A

5644: 87 357 :M

5645: .555 .055(impact of functional structures on predicting)J

5646: 70 371 :M

5647: .104 .01(the correct tree structures. We therefore removed)J

5648: 70 385 :M

5649: 2.037 .204(all f-structure units from )J

5650: 195 385 :M

5651: 2.106 .211(the fragments, thus)J

5652: 70 399 :M

5653: .819 .082(yielding a Tree-DOP )J

5654: 170 399 :M

5655: 1.312 .131(model, and compared the)J

5656: 70 413 :M

5657: .883 .088(results against the full LFG-DOP )J

5658: 228 413 :M

5659: 1.077 .108(model \(using)J

5660: 70 427 :M

5661: .122 .012(the discounted RF estimator and all fragments up)J

5662: 70 441 :M

5663: .515 .051(to depth 4\). We )J

5664: 145 441 :M

5665: .449 .045(evaluated the parse accuracy on)J

5666: 70 455 :M

5667: 3.264 .326(the tree structures only, using )J

5668: 230 455 :M

5669: 2.085 .208(exact match)J

5670: 70 469 :M

5671: 6.323 .632(together with the )J

5672: 178 469 :M

5673: 5.062 .506(standard PARSEVAL)J

5674: 70 483 :M

5675: 2.183 .218(measures. We used the same training/test )J

5676: 277 483 :M

5677: -.608(set)A

5678: 70 497 :M

5679: -.068(splits as in the previous experiments.)A

5680: 84 515 21 12 rC

5681: gS

5682: .781 .783 scale

5683: 120.781 665.834 :T

5684: -120.781 -665.834 :T

5685: 108.781 670.834 :M

5686: f3_12 sf

5687: (        )S

5688: gR

5689: gR

5690: gS 139 515 49 12 rC

5691: gS

5692: .781 .783 scale

5693: 209.169 665.834 :T

5694: -209.169 -665.834 :T

5695: 179.169 670.834 :M

5696: f3_12 sf

5697: (Exact Match)S

5698: gR

5699: gR

5700: gS 194 515 48 12 rC

5701: gS

5702: .781 .783 scale

5703: 279.556 665.834 :T

5704: -279.556 -665.834 :T

5705: 249.556 670.834 :M

5706: f3_12 sf

5707: (  Precision   )S

5708: gR

5709: gR

5710: gS 248 515 26 12 rC

5711: gS

5712: .781 .783 scale

5713: 333.664 665.834 :T

5714: -333.664 -665.834 :T

5715: 318.664 670.834 :M

5716: f3_12 sf

5717: (Recall)S

5718: gR

5719: gR

5720: gS 84 539 41 11 rC

5721: gS

5722: .781 .783 scale

5723: 133.781 695.223 :T

5724: -133.781 -695.223 :T

5725: 108.781 700.223 :M

5726: f3_12 sf

5727: (Tree-DOP)S

5728: gR

5729: gR

5730: gS 139 539 35 11 rC

5731: gS

5732: .781 .783 scale

5733: 200.169 695.223 :T

5734: -200.169 -695.223 :T

5735: 179.169 700.223 :M

5736: f3_12 sf

5737: (    46.6%)S

5738: gR

5739: gR

5740: gS 194 539 35 11 rC

5741: gS

5742: .781 .783 scale

5743: 270.556 695.223 :T

5744: -270.556 -695.223 :T

5745: 249.556 700.223 :M

5746: f3_12 sf

5747: (    88.9%)S

5748: gR

5749: gR

5750: gS 248 539 26 11 rC

5751: gS

5752: .781 .783 scale

5753: 333.664 695.223 :T

5754: -333.664 -695.223 :T

5755: 318.664 700.223 :M

5756: f3_12 sf

5757: (86.7%)S

5758: gR

5759: gR

5760: gS 84 562 42 12 rC

5761: gS

5762: .781 .783 scale

5763: 133.781 725.89 :T

5764: -133.781 -725.89 :T

5765: 108.781 730.89 :M

5766: f3_12 sf

5767: (LFG-DOP)S

5768: gR

5769: gR

5770: gS 139 562 35 12 rC

5771: gS

5772: .781 .783 scale

5773: 200.169 725.89 :T

5774: -200.169 -725.89 :T

5775: 179.169 730.89 :M

5776: f3_12 sf

5777: (    50.8%)S

5778: gR

5779: gR

5780: gS 194 562 80 12 rC

5781: gS

5782: .781 .783 scale

5783: 299.556 725.89 :T

5784: -299.556 -725.89 :T

5785: 249.556 730.89 :M

5786: f3_12 sf

5787: (    90.3%         88.4%)S

5788: gR

5789: gR

5790: gS 82 509 193 72 rC

5791: 84 533 -1 1 273 532 1 84 532 @a

5792: 84 512 -1 1 273 511 1 84 511 @a

5793: 84 581 -1 1 274 580 1 84 580 @a

5794: 90 518 26 9 rC

5795: gS

5796: .781 .783 scale

5797: 131.46 667.834 :T

5798: -131.46 -667.834 :T

5799: 116.46 670.834 :M

5800: f3_12 sf

5801: (Model)S

5802: gR

5803: gR

5804: gS 29 30 538 781 rC

5805: 90 601 :M

5806: f3_11 sf

5807: -.108(Table 5. Tree accuracy on the Verbmobil)A

5808: 86 624 20 12 rC

5809: gS

5810: .77 .761 scale

5811: 123.971 826.982 :T

5812: -123.971 -826.982 :T

5813: 112.971 831.982 :M

5814: f3_12 sf

5815: (        )S

5816: gR

5817: gR

5818: gS 140 624 49 12 rC

5819: gS

5820: .77 .761 scale

5821: 213.09 826.982 :T

5822: -213.09 -826.982 :T

5823: 183.09 831.982 :M

5824: f3_12 sf

5825: (Exact Match)S

5826: gR

5827: gR

5828: gS 194 624 48 12 rC

5829: gS

5830: .77 .761 scale

5831: 282.21 826.982 :T

5832: -282.21 -826.982 :T

5833: 253.21 831.982 :M

5834: f3_12 sf

5835: (  Precision   )S

5836: gR

5837: gR

5838: gS 248 624 25 12 rC

5839: gS

5840: .77 .761 scale

5841: 338.33 826.982 :T

5842: -338.33 -826.982 :T

5843: 323.33 831.982 :M

5844: f3_12 sf

5845: (Recall)S

5846: gR

5847: gR

5848: gS 86 647 40 11 rC

5849: gS

5850: .77 .761 scale

5851: 136.971 857.212 :T

5852: -136.971 -857.212 :T

5853: 112.971 862.212 :M

5854: f3_12 sf

5855: (Tree-DOP)S

5856: gR

5857: gR

5858: gS 140 647 35 11 rC

5859: gS

5860: .77 .761 scale

5861: 204.09 857.212 :T

5862: -204.09 -857.212 :T

5863: 183.09 862.212 :M

5864: f3_12 sf

5865: (    49.0%)S

5866: gR

5867: gR

5868: gS 194 647 35 11 rC

5869: gS

5870: .77 .761 scale

5871: 274.21 857.212 :T

5872: -274.21 -857.212 :T

5873: 253.21 862.212 :M

5874: f3_12 sf

5875: (    93.4%)S

5876: gR

5877: gR

5878: gS 248 647 25 11 rC

5879: gS

5880: .77 .761 scale

5881: 338.33 857.212 :T

5882: -338.33 -857.212 :T

5883: 323.33 862.212 :M

5884: f3_12 sf

5885: (92.1%)S

5886: gR

5887: gR

5888: gS 86 670 42 11 rC

5889: gS

5890: .77 .761 scale

5891: 137.971 887.442 :T

5892: -137.971 -887.442 :T

5893: 112.971 892.442 :M

5894: f3_12 sf

5895: (LFG-DOP)S

5896: gR

5897: gR

5898: gS 140 670 35 11 rC

5899: gS

5900: .77 .761 scale

5901: 204.09 887.442 :T

5902: -204.09 -887.442 :T

5903: 183.09 892.442 :M

5904: f3_12 sf

5905: (    53.2%)S

5906: gR

5907: gR

5908: gS 194 670 79 11 rC

5909: gS

5910: .77 .761 scale

5911: 303.21 887.442 :T

5912: -303.21 -887.442 :T

5913: 253.21 892.442 :M

5914: f3_12 sf

5915: (    95.8%         94.7%)S

5916: gR

5917: gR

5918: gS 83 618 191 70 rC

5919: 85 621 -1 1 271 620 1 85 620 @a

5920: 85 642 -1 1 272 641 1 85 641 @a

5921: 85 688 -1 1 272 687 1 85 687 @a

5922: 91 626 26 10 rC

5923: gS

5924: .77 .761 scale

5925: 134.463 828.982 :T

5926: -134.463 -828.982 :T

5927: 119.463 831.982 :M

5928: f3_12 sf

5929: (Model)S

5930: gR

5931: gR

5932: gS 29 30 538 781 rC

5933: 88 708 :M

5934: f3_11 sf

5935: -.133(Table 6. Tree accuracy on the Homecentre)A

5936: 70 731 :M

5937: 1.036 .104(The results indicate that LFG-DOP's functional)J

5938: 70 745 :M

5939: .105 .01(structures help to )J

5940: 150 745 :M

5941: .658 .066(improve the parse accuracy of)J

5942: 305 96 :M

5943: 4.009 .401(tree structures. In other words, )J

5944: 476 96 :M

5945: .094(LFG-DOP)A

5946: 305 110 :M

5947: 2.957 .296(outperforms Tree-DOP if evaluated )J

5948: 487 110 :M

5949: 2.98 .298(on tree)J

5950: 305 124 :M

5951: 1.119 .112(structures only. According )J

5952: 432 124 :M

5953: 1.884 .188(to paired )J

5954: 481 124 :M

5955: f4_11 sf

5956: (t)S

5957: 484 124 :M

5958: f3_11 sf

5959: 1.076 .108(-tests all)J

5960: 305 138 :M

5961: 5.524 .552(differences in accuracy )J

5962: 440 138 :M

5963: 3.255 .325(were statistically)J

5964: 305 152 :M

5965: .672 .067(significant. This result is )J

5966: 422 152 :M

5967: 1.064 .106(promising since Tree)J

5968: 520 152 :M

5969: (-)S

5970: 305 166 :M

5971: 1.733 .173(DOP )J

5972: 333 166 :M

5973: 1.191 .119(has been shown to obtain state-of-the-art)J

5974: 305 180 :M

5975: .863 .086(performance on )J

5976: 381 180 :M

5977: 1.425 .143(the Wall Street Journal corpus)J

5978: 305 194 :M

5979: -.055(\(see Bod 2000a\).)A

5980: 305 217 :M

5981: f0_11 sf

5982: -.021(5.4 Comparing Viterbi )A

5983: f6_11 sf

5984: (n)S

5985: 421 217 :M

5986: f0_11 sf

5987: .082 .008( best to Monte Carlo)J

5988: 305 231 :M

5989: f3_11 sf

5990: 1.858 .186(Finally, we )J

5991: 364 231 :M

5992: 2.272 .227(were interested in comparing an)J

5993: 305 245 :M

5994: 1.816 .182(alternative, more efficient )J

5995: 433 245 :M

5996: 2.944 .294(search method for)J

5997: 305 259 :M

5998: 1.924 .192(estimating the most probable analysis. )J

5999: 494 259 :M

6000: 2.434 .243(In the)J

6001: 305 273 :M

6002: .309 .031(following set of experiments )J

6003: 437 273 :M

6004: .979 .098(we use a Viterbi )J

6005: 518 273 :M

6006: f4_11 sf

6007: (n)S

6008: 305 287 :M

6009: f3_11 sf

6010: .662 .066(best search heuristic \(as explained in section 4\),)J

6011: 305 301 :M

6012: -.201(and let )A

6013: f4_11 sf

6014: (n)S

6015: 342 301 :M

6016: f3_11 sf

6017: .348 .035( range from 1 to 10,000 derivations. We)J

6018: 305 315 :M

6019: -.022(also compute the results obtained by Monte Carlo)A

6020: 305 329 :M

6021: .309 .031(for the same number of derivations. We used the)J

6022: 305 343 :M

6023: 1.404 .14(same training/test set splits as )J

6024: 451 343 :M

6025: 1.655 .166(in the previous)J

6026: 305 357 :M

6027: 2.099 .21(experiments and used both )J

6028: 440 357 :M

6029: .648 .065(ungeneralized and)J

6030: 305 371 :M

6031: -.121(generalized fragments up )A

6032: 417 371 :M

6033: .145 .014(to depth 4 together with)J

6034: 305 385 :M

6035: -.092(the discounted RF estimator.)A

6036: 306 404 65 12 rC

6037: gS

6038: .759 .742 scale

6039: 446.616 551.658 :T

6040: -446.616 -551.658 :T

6041: 404.616 556.658 :M

6042: f3_12 sf

6043: (Nr. of derivations)S

6044: gR

6045: gR

6046: gS 393 404 29 12 rC

6047: gS

6048: .759 .742 scale

6049: 536.28 551.658 :T

6050: -536.28 -551.658 :T

6051: 519.28 556.658 :M

6052: f3_12 sf

6053: (Viterbi )S

6054: gR

6055: gR

6056: gS 420 404 6 12 rC

6057: gS

6058: .759 .742 scale

6059: 556.865 551.658 :T

6060: -556.865 -551.658 :T

6061: 554.865 556.658 :M

6062: f4_12 sf

6063: (n)S

6064: gR

6065: gR

6066: gS 425 404 18 12 rC

6067: gS

6068: .759 .742 scale

6069: 571.455 551.658 :T

6070: -571.455 -551.658 :T

6071: 561.455 556.658 :M

6072: f3_12 sf

6073: ( best)S

6074: gR

6075: gR

6076: gS 473 404 47 12 rC

6077: gS

6078: .759 .742 scale

6079: 653.717 551.658 :T

6080: -653.717 -551.658 :T

6081: 624.717 556.658 :M

6082: f3_12 sf

6083: (Monte Carlo)S

6084: gR

6085: gR

6086: gS 325 427 7 11 rC

6087: gS

6088: .759 .742 scale

6089: 432.658 582.658 :T

6090: -432.658 -582.658 :T

6091: 429.658 587.658 :M

6092: f3_12 sf

6093: (1)S

6094: gR

6095: gR

6096: gS 405 427 25 11 rC

6097: gS

6098: .759 .742 scale

6099: 550.095 582.658 :T

6100: -550.095 -582.658 :T

6101: 535.095 587.658 :M

6102: f3_12 sf

6103: (74.8%)S

6104: gR

6105: gR

6106: gS 485 427 25 11 rC

6107: gS

6108: .759 .742 scale

6109: 655.533 582.658 :T

6110: -655.533 -582.658 :T

6111: 640.533 587.658 :M

6112: f3_12 sf

6113: (20.1%)S

6114: gR

6115: gR

6116: gS 325 438 11 11 rC

6117: gS

6118: .759 .742 scale

6119: 435.658 597.484 :T

6120: -435.658 -597.484 :T

6121: 429.658 602.484 :M

6122: f3_12 sf

6123: (10)S

6124: gR

6125: gR

6126: gS 405 438 25 11 rC

6127: gS

6128: .759 .742 scale

6129: 550.095 597.484 :T

6130: -550.095 -597.484 :T

6131: 535.095 602.484 :M

6132: f3_12 sf

6133: (75.3%)S

6134: gR

6135: gR

6136: gS 485 438 25 11 rC

6137: gS

6138: .759 .742 scale

6139: 655.533 597.484 :T

6140: -655.533 -597.484 :T

6141: 640.533 602.484 :M

6142: f3_12 sf

6143: (36.7%)S

6144: gR

6145: gR

6146: gS 325 449 16 11 rC

6147: gS

6148: .759 .742 scale

6149: 438.658 612.31 :T

6150: -438.658 -612.31 :T

6151: 429.658 617.31 :M

6152: f3_12 sf

6153: (100)S

6154: gR

6155: gR

6156: gS 405 449 25 11 rC

6157: gS

6158: .759 .742 scale

6159: 550.095 612.31 :T

6160: -550.095 -612.31 :T

6161: 535.095 617.31 :M

6162: f3_12 sf

6163: (77.5%)S

6164: gR

6165: gR

6166: gS 485 449 25 11 rC

6167: gS

6168: .759 .742 scale

6169: 655.533 612.31 :T

6170: -655.533 -612.31 :T

6171: 640.533 617.31 :M

6172: f3_12 sf

6173: (67.0%)S

6174: gR

6175: gR

6176: gS 325 460 22 11 rC

6177: gS

6178: .759 .742 scale

6179: 442.658 627.137 :T

6180: -442.658 -627.137 :T

6181: 429.658 632.137 :M

6182: f3_12 sf

6183: (1,000)S

6184: gR

6185: gR

6186: gS 405 460 25 11 rC

6187: gS

6188: .759 .742 scale

6189: 550.095 627.137 :T

6190: -550.095 -627.137 :T

6191: 535.095 632.137 :M

6192: f3_12 sf

6193: (77.5%)S

6194: gR

6195: gR

6196: gS 485 460 25 11 rC

6197: gS

6198: .759 .742 scale

6199: 655.533 627.137 :T

6200: -655.533 -627.137 :T

6201: 640.533 632.137 :M

6202: f3_12 sf

6203: (77.1%)S

6204: gR

6205: gR

6206: gS 325 471 27 11 rC

6207: gS

6208: .759 .742 scale

6209: 445.658 641.963 :T

6210: -445.658 -641.963 :T

6211: 429.658 646.963 :M

6212: f3_12 sf

6213: (10,000)S

6214: gR

6215: gR

6216: gS 405 471 25 11 rC

6217: gS

6218: .759 .742 scale

6219: 550.095 641.963 :T

6220: -550.095 -641.963 :T

6221: 535.095 646.963 :M

6222: f3_12 sf

6223: (77.5%)S

6224: gR

6225: gR

6226: gS 485 471 25 11 rC

6227: gS

6228: .759 .742 scale

6229: 655.533 641.963 :T

6230: -655.533 -641.963 :T

6231: 640.533 646.963 :M

6232: f3_12 sf

6233: (77.5%)S

6234: gR

6235: gR

6236: gS 305 397 217 92 rC

6237: 307 399 -1 1 521 398 1 307 398 @a

6238: 307 422 -1 1 521 421 1 307 421 @a

6239: 307 489 -1 1 521 488 1 307 488 @a

6240: gR

6241: gS 29 30 538 781 rC

6242: 335 509 :M

6243: f3_11 sf

6244: -.09(Table 7. Precision on the Verbmobil)A

6245: 306 533 66 12 rC

6246: gS

6247: .769 .742 scale

6248: 441.095 725.529 :T

6249: -441.095 -725.529 :T

6250: 399.095 730.529 :M

6251: f3_12 sf

6252: (Nr. of derivations)S

6253: gR

6254: gR

6255: gS 394 533 29 12 rC

6256: gS

6257: .769 .742 scale

6258: 530.493 725.529 :T

6259: -530.493 -725.529 :T

6260: 513.493 730.529 :M

6261: f3_12 sf

6262: (Viterbi )S

6263: gR

6264: gR

6265: gS 422 533 6 12 rC

6266: gS

6267: .769 .742 scale

6268: 551.892 725.529 :T

6269: -551.892 -725.529 :T

6270: 549.892 730.529 :M

6271: f4_12 sf

6272: (n)S

6273: gR

6274: gR

6275: gS 427 533 18 12 rC

6276: gS

6277: .769 .742 scale

6278: 565.092 725.529 :T

6279: -565.092 -725.529 :T

6280: 555.092 730.529 :M

6281: f3_12 sf

6282: ( best)S

6283: gR

6284: gR

6285: gS 476 533 47 12 rC

6286: gS

6287: .769 .742 scale

6288: 649.091 725.529 :T

6289: -649.091 -725.529 :T

6290: 620.091 730.529 :M

6291: f3_12 sf

6292: (Monte Carlo)S

6293: gR

6294: gR

6295: gS 305 526 220 92 rC

6296: 307 528 -1 1 524 527 1 307 527 @a

6297: 307 551 -1 1 524 550 1 307 550 @a

6298: 307 618 -1 1 524 617 1 307 617 @a

6299: 325 556 6 11 rC

6300: gS

6301: .769 .742 scale

6302: 425.794 756.529 :T

6303: -425.794 -756.529 :T

6304: 423.794 761.529 :M

6305: f3_12 sf

6306: (1)S

6307: gR

6308: gR

6309: gS 406 556 25 11 rC

6310: gS

6311: .769 .742 scale

6312: 544.093 756.529 :T

6313: -544.093 -756.529 :T

6314: 529.093 761.529 :M

6315: f3_12 sf

6316: (75.6%)S

6317: gR

6318: gR

6319: gS 487 556 25 11 rC

6320: gS

6321: .769 .742 scale

6322: 648.091 756.529 :T

6323: -648.091 -756.529 :T

6324: 633.091 761.529 :M

6325: f3_12 sf

6326: (25.6%)S

6327: gR

6328: gR

6329: gS 325 567 11 11 rC

6330: gS

6331: .769 .742 scale

6332: 428.794 771.356 :T

6333: -428.794 -771.356 :T

6334: 423.794 776.356 :M

6335: f3_12 sf

6336: (10)S

6337: gR

6338: gR

6339: gS 406 567 25 11 rC

6340: gS

6341: .769 .742 scale

6342: 544.093 771.356 :T

6343: -544.093 -771.356 :T

6344: 529.093 776.356 :M

6345: f3_12 sf

6346: (76.2%)S

6347: gR

6348: gR

6349: gS 487 567 25 11 rC

6350: gS

6351: .769 .742 scale

6352: 648.091 771.356 :T

6353: -648.091 -771.356 :T

6354: 633.091 776.356 :M

6355: f3_12 sf

6356: (44.3%)S

6357: gR

6358: gR

6359: gS 325 578 15 11 rC

6360: gS

6361: .769 .742 scale

6362: 431.794 786.182 :T

6363: -431.794 -786.182 :T

6364: 423.794 791.182 :M

6365: f3_12 sf

6366: (100)S

6367: gR

6368: gR

6369: gS 406 578 25 11 rC

6370: gS

6371: .769 .742 scale

6372: 544.093 786.182 :T

6373: -544.093 -786.182 :T

6374: 529.093 791.182 :M

6375: f3_12 sf

6376: (79.1%)S

6377: gR

6378: gR

6379: gS 487 578 25 11 rC

6380: gS

6381: .769 .742 scale

6382: 648.091 786.182 :T

6383: -648.091 -786.182 :T

6384: 633.091 791.182 :M

6385: f3_12 sf

6386: (74.6%)S

6387: gR

6388: gR

6389: gS 325 589 22 11 rC

6390: gS

6391: .769 .742 scale

6392: 436.794 801.008 :T

6393: -436.794 -801.008 :T

6394: 423.794 806.008 :M

6395: f3_12 sf

6396: (1,000)S

6397: gR

6398: gR

6399: gS 406 589 25 11 rC

6400: gS

6401: .769 .742 scale

6402: 544.093 801.008 :T

6403: -544.093 -801.008 :T

6404: 529.093 806.008 :M

6405: f3_12 sf

6406: (79.8%)S

6407: gR

6408: gR

6409: gS 487 589 25 11 rC

6410: gS

6411: .769 .742 scale

6412: 648.091 801.008 :T

6413: -648.091 -801.008 :T

6414: 633.091 806.008 :M

6415: f3_12 sf

6416: (79.1%)S

6417: gR

6418: gR

6419: gS 325 600 27 11 rC

6420: gS

6421: .769 .742 scale

6422: 439.794 815.834 :T

6423: -439.794 -815.834 :T

6424: 423.794 820.834 :M

6425: f3_12 sf

6426: (10,000)S

6427: gR

6428: gR

6429: gS 406 600 25 11 rC

6430: gS

6431: .769 .742 scale

6432: 544.093 815.834 :T

6433: -544.093 -815.834 :T

6434: 529.093 820.834 :M

6435: f3_12 sf

6436: (79.8%)S

6437: gR

6438: gR

6439: gS 487 600 25 11 rC

6440: gS

6441: .769 .742 scale

6442: 648.091 815.834 :T

6443: -648.091 -815.834 :T

6444: 633.091 820.834 :M

6445: f3_12 sf

6446: (80.0%)S

6447: gR

6448: gR

6449: gS 29 30 538 781 rC

6450: 332 638 :M

6451: f3_11 sf

6452: -.09(Table 8. Precision on the Homecentre)A

6453: 305 661 :M

6454: 1.653 .165(The )J

6455: 328 661 :M

6456: 2.575 .257(tables show that Viterbi )J

6457: 454 661 :M

6458: f4_11 sf

6459: .623(n)A

6460: f3_11 sf

6461: 1.968 .197( best already)J

6462: 305 675 :M

6463: .28 .028(achieves a maximum accuracy at 100 )J

6464: 476 675 :M

6465: -.187(derivations)A

6466: 305 689 :M

6467: -.028(\(at least )A

6468: 343 689 :M

6469: .826 .083(on the Verbmobil corpus\) while Monte)J

6470: 305 703 :M

6471: -.237(Carlo )A

6472: 332 703 :M

6473: .286 .029(needs a much larger number of derivations)J

6474: 305 717 :M

6475: 2.569 .257(to obtain )J

6476: 355 717 :M

6477: 3.002 .3(these results. On the Homecentre)J

6478: 305 731 :M

6479: -.117(corpus, )A

6480: 340 731 :M

6481: .119 .012(Monte Carlo slightly outperforms Viterbi)J

6482: 305 745 :M

6483: f4_11 sf

6484: (n)S

6485: 311 745 :M

6486: f3_11 sf

6487: -.006( best at 10,000 derivations, but these differences)A

6488: endp

6489: %%Page: 8 8

6490: %%BeginPageSetup

6491: initializepage

6492: (rens; page: 8 of 8)setjob

6493: %%EndPageSetup

6494: -29 -30 :T

6495: gS 29 30 538 781 rC

6496: 70 96 :M

6497: f3_11 sf

6498: .748 .075(are not )J

6499: 106 96 :M

6500: .456 .046(statistically significant. Also remarkable)J

6501: 70 110 :M

6502: -.125(are the relatively high results obtained with Viterbi)A

6503: 70 124 :M

6504: f4_11 sf

6505: (n)S

6506: 76 124 :M

6507: f3_11 sf

6508: .379 .038( best if only )J

6509: 134 124 :M

6510: .816 .082(one derivation is used. This score)J

6511: 70 138 :M

6512: -.008(corresponds to the analysis generated by the most)A

6513: 70 152 :M

6514: -.143(probable \(valid\) )A

6515: 142 152 :M

6516: -.009(derivation. Thus Viterbi )A

6517: f4_11 sf

6518: (n)S

6519: 257 152 :M

6520: f3_11 sf

6521: .537 .054( best is)J

6522: 70 166 :M

6523: .326 .033(a promising )J

6524: 127 166 :M

6525: .318 .032(alternative to Monte Carlo resulting)J

6526: 70 180 :M

6527: -.047(in a speed up of about two orders of magnitude.)A

6528: 70 204 :M

6529: f0_12 sf

6530: .207 .021(6  Conclusion)J

6531: 70 218 :M

6532: f3_11 sf

6533: .321 .032(We presented a parser which analyzes new )J

6534: 266 218 :M

6535: -.153(input)A

6536: 70 232 :M

6537: 1.875 .187(by )J

6538: 87 232 :M

6539: 1.249 .125(probabilistically combining fragments from)J

6540: 70 246 :M

6541: 1.255 .125(LFG-annotated corpora into new )J

6542: 227 246 :M

6543: 1.417 .142(analyses. We)J

6544: 70 260 :M

6545: .355 .036(have seen that the parse accuracy increased with)J

6546: 70 274 :M

6547: 3.488 .349(increasing fragment )J

6548: 175 274 :M

6549: 4.998 .5(size, and that LFG's)J

6550: 70 288 :M

6551: 1.26 .126(functional structures contribute )J

6552: 219 288 :M

6553: .684 .068(to significantly)J

6554: 70 302 :M

6555: 2.221 .222(higher parse accuracy on tree structures. We)J

6556: 70 316 :M

6557: 3.3 .33(tested two search )J

6558: 165 316 :M

6559: 3.666 .367(techniques for the most)J

6560: 70 330 :M

6561: -.118(probable analysis, Viterbi )A

6562: 184 330 :M

6563: f4_11 sf

6564: -.205(n)A

6565: f3_11 sf

6566: -.19( best)A

6567: 210 330 :M

6568: f4_11 sf

6569: -.096( )A

6570: f3_11 sf

6571: -.195(and Monte )A

6572: 262 330 :M

6573: -.237(Carlo.)A

6574: 70 344 :M

6575: 1.108 .111(While these two techniques achieved about the)J

6576: 70 358 :M

6577: 1.188 .119(same accuracy, Viterbi )J

6578: 181 358 :M

6579: f4_11 sf

6580: .819(n)A

6581: f3_11 sf

6582: 2.228 .223( best was about 100)J

6583: 70 372 :M

6584: -.072(times faster than Monte Carlo.)A

6585: 70 396 :M

6586: f0_12 sf

6587: -.106(References)A

6588: 70 409 :M

6589: f3_10 sf

6590: 6.444 .644(E.\312Black et al., 1991. "A Procedure for)J

6591: 89 422 :M

6592: 5.952 .595(Quantitatively Comparing the Syntactic)J

6593: 89 435 :M

6594: 2.361 .236(Coverage of English", )J

6595: f4_10 sf

6596: 4.352 .435(Proceedings DARPA)J

6597: 89 448 :M

6598: .202(Workshop)A

6599: f3_10 sf

6600: .797 .08(, Pacific Grove, Morgan Kaufmann.)J

6601: 70 461 :M

6602: 3.701 .37(R.\312Bod, 1993. "Using an Annotated Language)J

6603: 89 474 :M

6604: 3.431 .343(Corpus as a Virtual Stochastic Grammar",)J

6605: 89 487 :M

6606: f4_10 sf

6607: .557 .056(Proceedings AAAI'93)J

6608: f3_10 sf

6609: .389 .039(, Washington D.C.)J

6610: 70 500 :M

6611: 1.734 .173(R.\312Bod, 1998. )J

6612: f4_10 sf

6613: 2.392 .239(Beyond Grammar: An Experience-)J

6614: 89 513 :M

6615: 1.048 .105(Based Theory of Language,)J

6616: f3_10 sf

6617: 1.078 .108( CSLI Publications,)J

6618: 89 526 :M

6619: .816 .082(Cambridge University Press.)J

6620: 70 539 :M

6621: 6.851 .685(R.\312Bod 1999. "Context-Sensitive Dialogue)J

6622: 89 552 :M

6623: 2.926 .293(Processing with the DOP Model", )J

6624: f4_10 sf

6625: .899(Natural)A

6626: 89 565 :M

6627: .825 .082(Language Engineering)J

6628: f3_10 sf

6629: .475 .047( 5\(4\), 309-323.)J

6630: 70 578 :M

6631: 5.654 .565(R.\312Bod, 2000a. "Parsing with the Shortest)J

6632: 89 591 :M

6633: 4.349 .435(Derivation", )J

6634: f4_10 sf

6635: 6.577 .658(Proceedings COLING-2000)J

6636: f3_10 sf

6637: (,)S

6638: 89 604 :M

6639: 1.11 .111(Saarbr\237cken, Germany.)J

6640: 70 617 :M

6641: 1.221 .122(R.\312Bod 2000b. "Combining Semantic and Syntactic)J

6642: 89 630 :M

6643: 2.063 .206(Structure for Language Modeling", )J

6644: f4_10 sf

6645: .558(Proceed-)A

6646: 89 643 :M

6647: .822 .082(ings ICSLP-2000)J

6648: f3_10 sf

6649: .543 .054(, Beijing, China.)J

6650: 70 656 :M

6651: 1.727 .173(R.\312Bod 2000c. "An Empirical Evaluation of LFG-)J

6652: 89 669 :M

6653: 3.399 .34(DOP", )J

6654: f4_10 sf

6655: 6.375 .638(Proceedings COLING-2000)J

6656: f3_10 sf

6657: 3.05 .305(, Saar-)J

6658: 89 682 :M

6659: .805 .081(br\237cken, Germany.)J

6660: 70 695 :M

6661: 1.857 .186(R.\312Bod, 2000d. "The Storage and Computation of)J

6662: 89 708 :M

6663: 2.897 .29(Frequent Sentences", )J

6664: f4_10 sf

6665: 4.445 .445(Proceedings AMLAP-)J

6666: 89 721 :M

6667: .212(2000)A

6668: f3_10 sf

6669: .824 .082(, Leiden, The Netherlands.)J

6670: 70 734 :M

6671: 3.43 .343(R.\312Bod and R. Kaplan, 1998. "A Probabilistic)J

6672: 89 747 :M

6673: 2.285 .229(Corpus-Driven Model for Lexical Functional)J

6674: 324 95 :M

6675: 3.649 .365(Analysis", )J

6676: f4_10 sf

6677: 6.108 .611(Proceedings COLING-ACL'98)J

6678: f3_10 sf

6679: (,)S

6680: 324 108 :M

6681: 1.486 .149(Montreal, Canada.)J

6682: 305 121 :M

6683: 1.652 .165(R.\312Bonnema, R. Bod and R. Scha, 1997. "A DOP)J

6684: 324 134 :M

6685: 8.678 .868(Model for Semantic Interpretation",)J

6686: 324 147 :M

6687: f4_10 sf

6688: 1.135 .113(Proceedings ACL/EACL-97)J

6689: f3_10 sf

6690: .622 .062(, Madrid, Spain.)J

6691: 305 160 :M

6692: 1.501 .15(J.\312Chappelier and M. Rajman, 2000. "Monte Carlo)J

6693: 324 173 :M

6694: 1.391 .139(Sampling for NP-hard Maximization Problems)J

6695: 324 186 :M

6696: 2.636 .264(in the Framework of Weighted Parsing", in)J

6697: 324 199 :M

6698: f4_10 sf

6699: 5.635 .563(NLP 2000, Lecture Notes in Artificial)J

6700: 324 212 :M

6701: .578 .058(Intelligence 1835)J

6702: f3_10 sf

6703: .482 .048(, 106-117.)J

6704: 305 225 :M

6705: 2.304 .23(B.\312Cormons, 1999. )J

6706: f4_10 sf

6707: 2.8 .28(Analyse et d\216sambiguisation:)J

6708: 324 238 :M

6709: .718 .072(Une approche \210 base de corpus \(Data-Oriented)J

6710: 324 251 :M

6711: 3.272 .327(Parsing\) pour les r\216presentations lexicales)J

6712: 324 264 :M

6713: .893(fonctionnelles)A

6714: f3_10 sf

6715: 3.601 .36(. PhD thesis, Universit\216 de)J

6716: 324 277 :M

6717: 1.609 .161(Rennes, France.)J

6718: 305 290 :M

6719: 3.23 .323(I.\312Good, 1953. "The Population Frequencies of)J

6720: 324 303 :M

6721: 3.856 .386(Species and the Estimation of Population)J

6722: 324 316 :M

6723: .884 .088(Parameters", )J

6724: f4_10 sf

6725: .169(Biometrika)A

6726: f3_10 sf

6727: .633 .063( 40, 237-264.)J

6728: 305 329 :M

6729: 1.222 .122(J.\312Goodman, 1998. )J

6730: f4_10 sf

6731: 1.629 .163(Parsing Inside-Out)J

6732: f3_10 sf

6733: .936 .094(, PhD thesis,)J

6734: 324 342 :M

6735: .667 .067(Harvard University, Mass.)J

6736: 305 355 :M

6737: 4.099 .41(L.\312Hoogweg, 2000. )J

6738: f4_10 sf

6739: 3.838 .384(Enriching DOP1 with the)J

6740: 324 368 :M

6741: 4.696 .47(Insertion Operation)J

6742: f3_10 sf

6743: 2.528 .253(, MSc Thesis, Dept. of)J

6744: 324 381 :M

6745: .955 .095(Computer Science, University of Amsterdam.)J

6746: 305 394 :M

6747: 5.107 .511(R.\312Kaplan, and J. Bresnan, 1982. "Lexical-)J

6748: 324 407 :M

6749: 2.891 .289(Functional Grammar: A Formal System for)J

6750: 324 420 :M

6751: 2.879 .288(Grammatical Representation", in J. Bresnan)J

6752: 324 433 :M

6753: 5.471 .547(\(ed.\), )J

6754: f4_10 sf

6755: 8.24 .824(The Mental Representation of)J

6756: 324 446 :M

6757: 7.11 .711(Grammatical Relations)J

6758: f3_10 sf

6759: 3.531 .353(, The MIT Press,)J

6760: 324 459 :M

6761: 1.185 .119(Cambridge, Mass.)J

6762: 305 472 :M

6763: 2.151 .215(J.\312Maxwell and R. Kaplan, 1991. "A Method for)J

6764: 324 485 :M

6765: 3.45 .345(Disjunctive Constraint Satisfaction", in M.)J

6766: 324 498 :M

6767: 4.193 .419(Tomita \(ed.\), )J

6768: f4_10 sf

6769: 4.949 .495(Current Issues in Parsing)J

6770: 324 511 :M

6771: .209(Technology)A

6772: f3_10 sf

6773: .963 .096(, Kluwer Academic Publishers.)J

6774: 305 524 :M

6775: 4.4 .44(G.\312Neumann, 1998. "Automatic Extraction of)J

6776: 324 537 :M

6777: 2.665 .266(Stochastic Lexicalized Tree Grammars from)J

6778: 324 550 :M

6779: 1.877 .188(Treebanks", )J

6780: f4_10 sf

6781: 1.55 .155(Proceedings of the 4th Workshop)J

6782: 324 563 :M

6783: 3.771 .377(on Tree-Adjoining Grammars and Related)J

6784: 324 576 :M

6785: .249(Frameworks)A

6786: f3_10 sf

6787: .975 .097(, Philadelphia, PA.)J

6788: 305 589 :M

6789: 2.551 .255(G.\312Neumann and D. Flickinger, 1999. "Learning)J

6790: 324 602 :M

6791: 2.665 .266(Stochastic Lexicalized Tree Grammars from)J

6792: 324 615 :M

6793: 1.474 .147(HPSG", DFKI Technical Report, Saarbr\237cken,)J

6794: 324 628 :M

6795: .043(Germany.)A

6796: 305 641 :M

6797: 1.164 .116(H.\312Ney, S. Martin and F. Wessel, 1997. "Statistical)J

6798: 324 654 :M

6799: 1.665 .166(Language Modeling Using Leaving-One-Out",)J

6800: 324 667 :M

6801: 1.53 .153(in S. Young & G. Bloothooft \(eds.\), )J

6802: f4_10 sf

6803: .605(Corpus-)A

6804: 324 680 :M

6805: 4.299 .43(Based Methods in Language and Speech)J

6806: 324 693 :M

6807: .181(Processing)A

6808: f3_10 sf

6809: .877 .088(, Kluwer Academic Publishers.)J

6810: 305 706 :M

6811: 2.102 .21(K.\312Sima'an, 1999. )J

6812: f4_10 sf

6813: 2.76 .276(Learning Efficient Disambigu-)J

6814: 324 719 :M

6815: .71(ation)A

6816: f3_10 sf

6817: 2.962 .296(. PhD thesis, ILLC dissertation series)J

6818: 324 732 :M

6819: .496 .05(number 1999-02. Utrecht / Amsterdam.)J

6820: endp

6821: %%Trailer

6822: end

6823: %%EOF

6824: