File size: 79,636 Bytes
6fa4bc9 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 | {
"paper_id": "L16-1019",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T12:08:28.885560Z"
},
"title": "PentoRef: A Corpus of Spoken References in Task-oriented Dialogues",
"authors": [
{
"first": "Sina",
"middle": [],
"last": "Zarrie\u00df",
"suffix": "",
"affiliation": {
"laboratory": "Dialogue Systems Group",
"institution": "Bielefeld University",
"location": {}
},
"email": ""
},
{
"first": "Julian",
"middle": [],
"last": "Hough",
"suffix": "",
"affiliation": {
"laboratory": "Dialogue Systems Group",
"institution": "Bielefeld University",
"location": {}
},
"email": ""
},
{
"first": "Casey",
"middle": [],
"last": "Kennington",
"suffix": "",
"affiliation": {
"laboratory": "Dialogue Systems Group",
"institution": "Bielefeld University",
"location": {}
},
"email": ""
},
{
"first": "Ramesh",
"middle": [],
"last": "Manuvinakurike",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "USC Institute for Creative Technologies",
"location": {
"settlement": "Playa Vista",
"region": "CA"
}
},
"email": ""
},
{
"first": "David",
"middle": [],
"last": "Devault",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "USC Institute for Creative Technologies",
"location": {
"settlement": "Playa Vista",
"region": "CA"
}
},
"email": ""
},
{
"first": "Raquel",
"middle": [],
"last": "Fern\u00e1ndez",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "University of Amsterdam",
"location": {}
},
"email": ""
},
{
"first": "David",
"middle": [],
"last": "Schlangen",
"suffix": "",
"affiliation": {
"laboratory": "Dialogue Systems Group",
"institution": "Bielefeld University",
"location": {}
},
"email": ""
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "PentoRef is a corpus of task-oriented dialogues collected in systematically manipulated settings. The corpus is multilingual, with English and German sections, and overall comprises more than 20000 utterances. The dialogues are fully transcribed and annotated with referring expressions mapped to objects in corresponding visual scenes, which makes the corpus a rich resource for research on spoken referring expressions in generation and resolution. The corpus includes several sub-corpora that correspond to different dialogue situations where parameters related to interactivity, visual access, and verbal channel have been manipulated in systematic ways. The corpus thus lends itself to very targeted studies of reference in spontaneous dialogue.",
"pdf_parse": {
"paper_id": "L16-1019",
"_pdf_hash": "",
"abstract": [
{
"text": "PentoRef is a corpus of task-oriented dialogues collected in systematically manipulated settings. The corpus is multilingual, with English and German sections, and overall comprises more than 20000 utterances. The dialogues are fully transcribed and annotated with referring expressions mapped to objects in corresponding visual scenes, which makes the corpus a rich resource for research on spoken referring expressions in generation and resolution. The corpus includes several sub-corpora that correspond to different dialogue situations where parameters related to interactivity, visual access, and verbal channel have been manipulated in systematic ways. The corpus thus lends itself to very targeted studies of reference in spontaneous dialogue.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "We present PentoRef, a corpus of task-oriented spoken dialogues recorded in a puzzle-playing domain where players have to manipulate and communicate about Pentomino pieces. 1 PentoRef presents a rich resource for investigating human conversational strategies for referring to objects, on different levels of linguistic realization (including speech and timing/turn-taking) and in different yet consistently represented interactive and visual contexts. In particular, PentoRef is useful for developing automatic systems for, and studying the human mechanisms for, two concrete tasks, namely reference resolution (RR) and referring expression generation (REG). The corpus is a meta-collection that bundles up a range of experimental data collected over recent years in the Dialogue Systems Group, first at Potsdam University and then Bielefeld University, and by collaborators. The individual sub-corpora have been used for empirical studies of conversational behaviour in spoken language interaction as well as work on building statistical reference resolution systems in situated environments, in German and English (Fern\u00e1ndez et al., 2006; Schlangen and Fern\u00e1ndez, 2007; Schlangen et al., 2009; Heintze et al., 2010; Kennington et al., 2013; Kennington and Schlangen, 2015) . The common property of the experiments in this collection is that participants have to produce spoken referring expressions to puzzle pieces in a game, normally to instruct another player to carry out a certain move on the Pentomino game board. At the same time, some important parameters of the respective experimental settings were manipulated, such as the way communication was mediated (speech channel and/or visual channel), and the presentation of the scene (virtual or real-world). The original versions of the sub-corpora could not be directly exploited for systematic studies of referring expressions across these settings, due to inconsistent conventions used for segmenting, transcribing and annotating the audio recordings. More- 1 Pentomino is a puzzle game with pieces based on the 12 different shapes that can be constructed from arranging 5 squares next to each other. over, in each experiment, the visual scenes and visual attributes of pieces in a scene were represented in different ways (e.g. either as sets of logical properties or as low-level features from machine vision) such that additional annotation and standardization is needed to exploit the data as an actual corpus of spoken references. This paper presents the upcoming inaugural release of Pen-toRef, a unification of these resources that contains highquality transcriptions of spoken utterances, consistent representations of visual scenes, mark-up of referring expressions and mappings between referring expressions and pieces present in a visual scene. In addition to a consistently structured resource of the raw and derived data, we also provide a light-weight relational database that can be easily processed and queried across the different experimental settings in PentoRef.",
"cite_spans": [
{
"start": 173,
"end": 174,
"text": "1",
"ref_id": null
},
{
"start": 1116,
"end": 1140,
"text": "(Fern\u00e1ndez et al., 2006;",
"ref_id": "BIBREF3"
},
{
"start": 1141,
"end": 1171,
"text": "Schlangen and Fern\u00e1ndez, 2007;",
"ref_id": "BIBREF17"
},
{
"start": 1172,
"end": 1195,
"text": "Schlangen et al., 2009;",
"ref_id": "BIBREF18"
},
{
"start": 1196,
"end": 1217,
"text": "Heintze et al., 2010;",
"ref_id": "BIBREF7"
},
{
"start": 1218,
"end": 1242,
"text": "Kennington et al., 2013;",
"ref_id": "BIBREF11"
},
{
"start": 1243,
"end": 1274,
"text": "Kennington and Schlangen, 2015)",
"ref_id": "BIBREF10"
},
{
"start": 2019,
"end": 2020,
"text": "1",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1."
},
{
"text": "Compared to other resources used in dialogue research, PentoRef follows a tradition perhaps best exemplified by the HCRC Map Task Corpus (Anderson et al., 1991; MacMahon et al., 2006) in that it combines the naturalness of unscripted conversation with the advantages of taskoriented dialogue, such as careful control over aspects of the linguistic and extralinguistic context. Recent comparable data collection efforts are relatively rare, but see (Tokunaga et al., 2012; Gatt and Paggio, 2014) . Related studies in REG research showed that the linguistic phenomena found in the elicited referring expressions vary widely with the modality, task, and audience, cf. (Mitchell et al., 2010; Koolen and Krahmer, 2010; Clarke et al., 2013) . Inspired by a recently increasing interest in image description and labelling tasks, data sets of real-world photographs (paired with references to specific entities in the image) have also been created for REG (Kazemzadeh et al., 2014; Gkatzia et al., 2015) . Real-world images pose interesting challenges for REG, as the set of visual attributes and, consequently, the distractor objects (objects present in the scene which are not the target of a referring expression) cannot be directly controlled.",
"cite_spans": [
{
"start": 137,
"end": 160,
"text": "(Anderson et al., 1991;",
"ref_id": "BIBREF0"
},
{
"start": 161,
"end": 183,
"text": "MacMahon et al., 2006)",
"ref_id": "BIBREF14"
},
{
"start": 448,
"end": 471,
"text": "(Tokunaga et al., 2012;",
"ref_id": "BIBREF19"
},
{
"start": 472,
"end": 494,
"text": "Gatt and Paggio, 2014)",
"ref_id": "BIBREF5"
},
{
"start": 665,
"end": 688,
"text": "(Mitchell et al., 2010;",
"ref_id": "BIBREF16"
},
{
"start": 689,
"end": 714,
"text": "Koolen and Krahmer, 2010;",
"ref_id": "BIBREF12"
},
{
"start": 715,
"end": 735,
"text": "Clarke et al., 2013)",
"ref_id": "BIBREF2"
},
{
"start": 949,
"end": 974,
"text": "(Kazemzadeh et al., 2014;",
"ref_id": "BIBREF9"
},
{
"start": 975,
"end": 996,
"text": "Gkatzia et al., 2015)",
"ref_id": "BIBREF6"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2."
},
{
"text": "Although attempts have been made to systematically assess the effects of the different domains on the reference task (Gkatzia et al., 2015) , the comparability of existing reference corpora is limited as they are based on very different types of visual stimuli. PentoRef provides an unusually wide spectrum of experimental settings that have been investigated in a single domain, combining various levels of interactivity and mediation on the one hand, and variation between virtual and real-world scenes on the other.",
"cite_spans": [
{
"start": 117,
"end": 139,
"text": "(Gkatzia et al., 2015)",
"ref_id": "BIBREF6"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2."
},
{
"text": "PentoRef consists of different manipulations on taskoriented puzzle-playing using the 12 Pentomino pieces, individuated by their shape. When more than one set of Pentominoes is used, the object type may also be individuated by colour. An important difference to standard reference resources is that control over the set of distractors was not a major consideration during experiment design. Different settings vary widely with respect to number of pieces in a scene, and the properties that a target piece shares with distractor objects. For instance, in some settings, all pieces had the same color. In other settings, each piece had a unique color. Taken together as a corpus, the experiments thus provide an interesting test-bed for REG and RR systems that need to adapt to different types of visual contexts within a common domain.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "PentoRef Overview",
"sec_num": "3."
},
{
"text": "In the puzzle games, a player can have one of the following roles: (i) the Instruction Giver (IG), the player who has complete knowledge about the game's goal (e.g. a picture of a shape constructed out of Pentomino pieces), but who cannot manipulate the pieces herself, or (ii) the Instruction Follower (IF) who can manipulate pieces, but does not have knowledge about the game's goal. In order to achieve the goal, the IG has to formulate verbal instructions which the IF has to execute in terms of actions on the game board (i.e. selecting, moving, rotating, or placing pieces). In this task-oriented setting, it is possible to directly assess the communicative success (effectiveness) of an utterance or a referring expression in that if the IF could quickly identify the intended Pentomino piece in the scene, the referring expression formulated by the IG was immediately effective. In some of the interactions, only the piece selection is required of the IF rather than the construction of the entire puzzle, however reference identification is common to all domains. The corpus contains two main types of task-oriented interactions:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "General Task",
"sec_num": "3.1."
},
{
"text": "Human-wizard interaction: A human IG has the task to instruct what they believe to be a machine to select or move certain pieces on a game board or desk. Depending on the setting the IG can use speech, and sometimes, gesture. Behind the scenes, a human wizard performs the game actions as the IF. The IG receives signals of the wizard's game actions (e.g. via highlighted pieces on the screen, or audio signals). In some cases, the IG can react to these signals.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "General Task",
"sec_num": "3.1."
},
{
"text": "Human-human dialogues: The IF is a human player that communicates with the IG via speech. Both players collaboratively perform the task (i.e. building a shape out of Pentomino pieces). The IG has the desired solution to the puzzle, but cannot manipulate pieces, whereas the IG can manipulate pieces but does not have the solution. Table 1 shows an overview of the data that we have bundled up for PentoRef, and introduces the sub-corpora with their labels, as they were used in previous research. Experimental settings have been manipulated along the following dimensions.",
"cite_spans": [],
"ref_spans": [
{
"start": 331,
"end": 338,
"text": "Table 1",
"ref_id": "TABREF1"
}
],
"eq_spans": [],
"section": "General Task",
"sec_num": "3.1."
},
{
"text": "Scene: In virtual settings, Pentomino pieces are shown as graphical objects on a computer screen. In the realworld settings, participants had to interact with real pieces on a physical game board. There is also an intermediate level of \"images\" in the RDG-Pento experiment, a version of the RDG-Image game described in (Manuvinakurike et al., 2015) , using the same webbased data-collection methods using photographs of real Pentomino pieces.",
"cite_spans": [
{
"start": 319,
"end": 348,
"text": "(Manuvinakurike et al., 2015)",
"ref_id": "BIBREF15"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Experimental Settings",
"sec_num": "3.2."
},
{
"text": "Pre-solved game: When the game plan is pre-solved, the IG cannot decide on the pieces that the IF has to select and actions that the IF has to perform, but has to follow some plan given to them as a stimulus. When the game is not pre-solved, the IG can freely decide on the order of game actions, and potentially, the types of pieces the IF has to select.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Experimental Settings",
"sec_num": "3.2."
},
{
"text": "Vision: When vision is available, IGs can observe what the IF is doing, e.g. via a camera feed of the IF's game board and their hands, or the IF's mouse movements on a screen. Otherwise, participants only communicate via speech.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Experimental Settings",
"sec_num": "3.2."
},
{
"text": "In each experimental setting, players had to interact with Pentomino pieces. Beyond that common property, the different settings vary widely with respect to number of pieces in a scene, and the properties that a target piece would share with distractor objects. This is illustrated in Figure 1 , showing four example scenes from Take, Take-CV, Visual Pento, and WOz-Pento. For instance, in Visual Pento, all pieces initially have the same color (blue) and their shape uniquely distinguishes them from all other pieces. For the Take experiment, the scenes were randomly generated and contained a large number of pieces in various colors such that there were always pieces that had the same color and/or shape. As another example, the scenes in Take-CV were composed of real Pentomino pieces taken from 3 sets and randomly distributed on a desk. In this case, some colors only occur with a particular shape (e.g. red crosses). Moreover, there were wooden pieces or pieces with different shades of the same color. Another difference between the virtual and the real scenes concerns the orientation of the pieces. In the virtual scenes, the pieces were arranged on a regular rectangular grid. The ",
"cite_spans": [],
"ref_spans": [
{
"start": 285,
"end": 293,
"text": "Figure 1",
"ref_id": null
}
],
"eq_spans": [],
"section": "Scenes and Distractors",
"sec_num": "3.3."
},
{
"text": "Task In this Wizard-of-Oz study, users gave instructions to the system (the wizard) in order to manipulate (select, rotate, mirror, delete) puzzle pieces on an upper board and to put them onto a lower board, reaching a pre-specified goal state. Each participant took part in several rounds in which the distinguishing characteristics for puzzle pieces (color, shape, pro-posed name, position on the board) varied widely.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "WOz Pento",
"sec_num": "4.1."
},
{
"text": "Task In this Wizard-of-Oz study, the participant was confronted with a game board containing 15 randomly selected Pentomino puzzle pieces (out of a repertoire of 12 shapes, and 6 colors). The positions of the pieces were randomly determined, but in such a way that the pieces grouped in the four corners of the screen. They were instructed to (silently) choose a Pentomino tile on the screen and then instruct the computer system to select this piece by describing and pointing to it. When a piece was selected (by the wizard), the participant had to utter a confirmation (or give negative feedback) and a new board was generated and the process repeated.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Take",
"sec_num": "4.2."
},
{
"text": "Procedure The participants were seated at a table in front of the screen. Their gaze was then calibrated with an eye tracker (Seeingmachines FaceLab) placed above the screen and their arm movements (captured by a Microsoft Kinect, also above the screen) were also calibrated. The utterances, board states, arm movements, and gaze information were recorded in a similar fashion as described in (?). The wizard was instructed to elicit pointing gestures by waiting to select the participant-referred piece by several seconds, unless a pointing action by the participant had already occurred. When the wizard misunderstood, or a technical problem arose, the wizard had an option to flag the episode.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Take",
"sec_num": "4.2."
},
{
"text": "Task In this Wizard-of-Oz setting, participants were seated in front of a table with 36 Pentomino puzzle pieces that were randomly placed with some space between them.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Take-CV",
"sec_num": "4.3."
},
{
"text": "The task of the participant was to refer to that object using only speech, as if identifying it for a friend sitting next to the participant.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Take-CV",
"sec_num": "4.3."
},
{
"text": "Procedure Above the table was a camera that recorded a video feed of the objects, processed using OpenCV to segment the objects; of those, one (or one pair) was chosen randomly by the experiment software. The video image was presented to the participant on a display placed behind the table, but with the randomly selected piece (or pair of pieces) indicated by an overlay. The wizard had an identical screen depicting the scene but not the selected object. The wizard listened to the participants RE and clicked on the object she thought was being referred on her screen. If it was the target object, a tone sounded and a new object was randomly chosen. If a wrong object was clicked, a different tone sounded, the episode was flagged, and a new episode began. At varied intervals, the participant was instructed to \"shuffle\" the board between episodes by moving around the pieces.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Take-CV",
"sec_num": "4.3."
},
{
"text": "Phases The first half of the allotted time constituted Phase 1. After Phase 1 was complete, instructions for Phase 2 were explained: the screen showed the target and also a landmark object, outlined in blue, near the target. The participant was instructed to refer to the target using the landmark. (In the instructions, the concepts of landmark and target were explained in general terms.) All other instructions remained the same as Phase 1. The targets identifier, which was always known be-forehand, was always recorded. For Phase 2, the landmarks identifier was also recorded.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Take-CV",
"sec_num": "4.3."
},
{
"text": "Task The IG instructs the IF on how to build a Pentomino puzzle-an elephant shape built out of tiles that are composed out of five squares (see Figure 1) . The IG has the solution of the puzzle, while the IF is only given the outline and a set of 12 loose pieces. The Pentomino pieces available to the IF, while distinct in shape, are all the same colour and do not have an identifying label.",
"cite_spans": [],
"ref_spans": [
{
"start": 144,
"end": 153,
"text": "Figure 1)",
"ref_id": null
}
],
"eq_spans": [],
"section": "Noise/No-noise",
"sec_num": "4.4."
},
{
"text": "Conditions In Noise/No-Noise, there were two conditions: a Noise condition (experimental group) where the channel from the IG to the IF was manipulated by replacing, in real time and at random points, all signal with noise (brown noise, matched to loudness level of channel); and Figure 1 : A common reference mark-up across the PentoRef settings (the letters V and Z serve as shape identifiers)",
"cite_spans": [],
"ref_spans": [
{
"start": 280,
"end": 288,
"text": "Figure 1",
"ref_id": null
}
],
"eq_spans": [],
"section": "Noise/No-noise",
"sec_num": "4.4."
},
{
"text": "a No-noise condition (control group) where there were no manipulations.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Noise/No-noise",
"sec_num": "4.4."
},
{
"text": "Procedure Subjects were jointly greeted by the experimenter, who briefly explained the tasks to be carried out and allowed them to choose their roles as either IG or IF. They were then placed in different sound-proof rooms and were given written instructions for the Pentomino task. The IF was allowed a few minutes to get used to the Pentomino program. After subjects had read the instructions, the experimenter asked each of them whether they had any questions. Before leaving the IF room, the experimenter said to the IF something to the effect of: \"There might be some problems with the audio, which we can't fix at the moment, so please just go ahead\". This was done in order to prevent subjects in the noise condition from coming out of the room to complain about the quality of the audio. Finally the experimenter left the rooms and the first phase of the run began.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Noise/No-noise",
"sec_num": "4.4."
},
{
"text": "Task Same as Noise/No-noise.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Visual Pento",
"sec_num": "4.5."
},
{
"text": "Procedure The setting in this experiment was very much like the one described for the Pentomino task in the Noise experiment, except that there was a visual channel between IG and IF that allowes IG to see the actions performed by IF on the board. This was realised technically through a Virtual Network Computing (VNC) connection between the IF computer and a computer in IG's room, which replayed the GUI of the Pentomino program on which the IF was executing the instructions. Recording was done as described for the No-noise condition.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Visual Pento",
"sec_num": "4.5."
},
{
"text": "Task In this human-human set-up, two participants worked together to construct objects out of 12 pentomino tiles, one person could see the goal shape (the IG), the other could manipulate the objects (the IF). Each game was further subdivided into an initial selection phase and the actual game. In the selection phase, the IF picked some objects and presented them to the IG. The IG had to find a shape in a database with those objects. After that, the IG directed the IF in creating that shape.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Pento-CV",
"sec_num": "4.6."
},
{
"text": "Procedure Subjects were jointly greeted by the experimenter, who briefly explained the tasks to be carried out and allowed them to choose their initial roles as either IG or IF. They were then placed on different tables in the room. Above the table of the IF was a camera that recorded a video feed of the objects and his hands. The video image was presented to the IG on his screen. For each pair of participants, several games were recorded. After the first half of the allotted recording time, participants were asked to switch roles.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Pento-CV",
"sec_num": "4.6."
},
{
"text": "Task This is a Pentomino version of the Rapid Dialogue Game (RDG) described in (Manuvinakurike et al., 2015) , a human-human set-up where participants have audio access to each other through microphones and headsets. The participants had mutual visual access to a set of images, which are changed for each new round in the game. The participant playing the IG role would have one of the images on their screen highlighted as a target. They would describe the target to the participant in the IF role, who would try to identify it as fast as possible and click on the image they guessed to be the target. Participants were motivated by time pressure with the incentive to score as many points as possible in each fixed-duration round.",
"cite_spans": [
{
"start": 79,
"end": 108,
"text": "(Manuvinakurike et al., 2015)",
"ref_id": "BIBREF15"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "RDG-Pento",
"sec_num": "4.7."
},
{
"text": "Procedure Participants were recruited and their technical set-ups tested via the web in the way described by (Manuvinakurike et al., 2015) . Participants would follow on-screen instructions then begin their first round in one of the roles (IG or IF). In each round, the pairs were presented 8 images of Pentomino pieces at a time on their own screens. The participant roles were switched every round. There were several rounds per difficulty level, starting with the easiest task with images of single Pento pieces, then progressing to sets of 2-6 pieces in each image. See Figure 2 for an example of the level with 2 pieces per image.",
"cite_spans": [
{
"start": 109,
"end": 138,
"text": "(Manuvinakurike et al., 2015)",
"ref_id": "BIBREF15"
}
],
"ref_spans": [
{
"start": 574,
"end": 580,
"text": "Figure",
"ref_id": null
}
],
"eq_spans": [],
"section": "RDG-Pento",
"sec_num": "4.7."
},
{
"text": "PentoRef consists of recordings of spontaneous speech. Most REG corpora have been collected in written, noninteractive domains. However, it is well-known that when humans use referring expressions in more natural, interac- Figure 2 : Game board in the RDG setting tive and situated contexts, conversational strategies are entirely different (Clark and Krych, 2004) . Importantly, in a situated dialogue, conversation partners typically collaborate to identify a particular target object, often coordinating on a referring strategy. While the IG utters the RE, the listener (the IF) can give feedback signals (verbal or action-based), or ask for clarifications and engage in repair sequences. A frequent phenomenon is 'reference in installments' where speakers split the reference across several utterances to incrementally build common ground with the listener. On the other hand, in spoken interactions, speakers (instruction givers) do not have unlimited time to ponder an optimal RE to refer to a particular object in a potentially complex scene. As a result, spoken referring expressions (as spoken language in general) typically contain disfluent material, including interruptions, pauses, hesitations, repetitions and self-repairs. To illustrate that PentoRef captures these types of referring, we present a few examples. Example (1) taken from Visual Pento (cf. Figure 1) illustrates typical phenomena in spoken referring expressions, such as repair, interruption and hesitation.",
"cite_spans": [
{
"start": 341,
"end": 364,
"text": "(Clark and Krych, 2004)",
"ref_id": "BIBREF1"
}
],
"ref_spans": [
{
"start": 223,
"end": 231,
"text": "Figure 2",
"ref_id": null
},
{
"start": 1369,
"end": 1378,
"text": "Figure 1)",
"ref_id": null
}
],
"eq_spans": [],
"section": "Referring in Spoken Dialogue: Examples",
"sec_num": "5."
},
{
"text": "(1) a. IG: IG: In Example (1), the IG first uses an analogical expression to refer to a piece. This is misunderstood by the IF who does not select the intended referent. The IG immediately produces utterances that correct the IF's action and provides more information about the target. In Example (2), again taken from Visual Pento, the IG is not certain how to name the properties of the target piece in an optimal way (i.e. shape or location) so he uses the location of the mouse pointer as a landmark, and produces a hesitation, and a hypernym. The IF interrupts him and asks for feedback about his current piece selection. Ecke.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Referring in Spoken Dialogue: Examples",
"sec_num": "5."
},
{
"text": "Jetzt",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Referring in Spoken Dialogue: Examples",
"sec_num": "5."
},
{
"text": "In the RDG Pento data, participants had to refer to sets instead of individual Pentomino pieces. The following Example illustrates a referring expression from that sub-corpus (produced for the second set in the bottom row in Figure 2 ).",
"cite_spans": [],
"ref_spans": [
{
"start": 225,
"end": 234,
"text": "Figure 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "Referring in Spoken Dialogue: Examples",
"sec_num": "5."
},
{
"text": "(4) blue L on the top and the harry potter sign on the right Finally, we want to point out that our corpus also contains references to locations and a restricted set of actions. In the following example, taken from Pento-CV, the IG tries to explain to the IF how to position and rotate the object on the game board. As this example illustrates, this data is rich in disfluencies which are marked up according to the transcription and segmentation guidelines developed by (Hough et al., 2015 ",
"cite_spans": [
{
"start": 471,
"end": 490,
"text": "(Hough et al., 2015",
"ref_id": "BIBREF8"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Referring in Spoken Dialogue: Examples",
"sec_num": "5."
},
{
"text": "Here we briefly describe the representations we provide in the corpus. The available annotations and overall corpus statistics including word types and tokens in each experimental setting are summarized in Table 2 .",
"cite_spans": [],
"ref_spans": [
{
"start": 206,
"end": 213,
"text": "Table 2",
"ref_id": "TABREF7"
}
],
"eq_spans": [],
"section": "Data Representation for Dialogue, Scenes and References",
"sec_num": "6."
},
{
"text": "We provide high quality utterance segmentation and transcription according to the manual in (Hough et al., 2015) , all of which was quality checked by the first two au- ",
"cite_spans": [
{
"start": 92,
"end": 112,
"text": "(Hough et al., 2015)",
"ref_id": "BIBREF8"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Transcription and Segmentation",
"sec_num": "6.1."
},
{
"text": "Across all datasets we provide a common mark-up for objects, whereby each puzzle piece in a game has a unique ID. Also common across every setting are the two high-level attributes of piece shape 2 and colour from a closed set which is sufficient to identify all piece types across all settings. All referring expressions to pieces are marked with this identifying information over word spans. See Figure 1 which shows the commonality of this mark-up between the virtual and real-world settings. The reference annotation links the transcribed utterances to unique identifiers of pieces in the corresponding scene. In Take-CV, at the time of writing is the only corpus with landmark referents and relations such as 'next to' to be annotated in addition to the target referring expression.",
"cite_spans": [],
"ref_spans": [
{
"start": 398,
"end": 406,
"text": "Figure 1",
"ref_id": null
}
],
"eq_spans": [],
"section": "Referent and Scene Representation",
"sec_num": "6.2."
},
{
"text": "Visual Information from Scenes For RR and REG automatic tasks, one wishes to identify a referent in a scene given a representation of the scene and the words, so we make available both logical features and, for the real-world scenes, automatically derived real-valued machine vision captured features of each object in the scene. For example in Figure 1 , while the Take dataset provides logical features for a piece such as colour=red, in Take-CV, the features provided are from machine vision and will provide features such as RGB value, hue and saturation.",
"cite_spans": [],
"ref_spans": [
{
"start": 345,
"end": 353,
"text": "Figure 1",
"ref_id": null
}
],
"eq_spans": [],
"section": "Referent and Scene Representation",
"sec_num": "6.2."
},
{
"text": "2 Each object shape name is the letter that corresponds most closely to its shape in its normal orientation. Lightweight database Our data therefore represents the following layers of information: (i) transcribed words, (ii) segmentation of sequences of words into utterances, (iii) annotation of referring expression on word spans, (iv) representations of visual scenes. We use a light-weight relational database format to represent the data in PentoRef, shown in Figure 3 . Information on words, utterances and scenes are kept in tables that can be linked via the identifiers for pieces and referring expressions. Therefore, it is straightforward to query the database for all expressions referring to pieces with a particular shape across the different sub-corpora. In the general case, the scenes in our experiments are dynamic. This means that the location of pieces and their orientation on the game board changes over time. We include timestamps as unique identifiers for scenes.",
"cite_spans": [],
"ref_spans": [
{
"start": 465,
"end": 473,
"text": "Figure 3",
"ref_id": "FIGREF1"
}
],
"eq_spans": [],
"section": "Referent and Scene Representation",
"sec_num": "6.2."
},
{
"text": "PentoRef transcriptions and annotations are made available under a public PDDL license (doi:10.4119/unibi/ 2901444). Please contact the authors for obtaining audio data.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Release",
"sec_num": "7."
},
{
"text": "We have presented PentoRef, a spoken dialogue corpus consisting of several sub-corpora collected in systematically manipulated settings. The corpus includes a variety of dialogue situations that differ systematically with respect to interactivity, verbal channel, and visual access, which allows for interesting comparisons between experimental settings. The corpus is fully transcribed and enriched with different representations of visual scenes and annotations of referring expressions, providing a rich resource for reference in spontaneous spoken language.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "8."
}
],
"back_matter": [
{
"text": "This work was supported by the German Research Foundation (DFG) through the Cluster of Excellence Cognitive Interaction Technology 'CITEC' (EXC 277) at Bielefeld University and the DUEL project (grant SCHL 845/5-1).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Acknowledgements",
"sec_num": "9."
}
],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "The HCRC Map Task corpus",
"authors": [
{
"first": "A",
"middle": [
"H"
],
"last": "Anderson",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Bader",
"suffix": ""
},
{
"first": "E",
"middle": [
"G"
],
"last": "Bard",
"suffix": ""
},
{
"first": "E",
"middle": [],
"last": "Boyle",
"suffix": ""
},
{
"first": "G",
"middle": [],
"last": "Doherty",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Garrod",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Isard",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Kowtko",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Mcallister",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Miller",
"suffix": ""
}
],
"year": 1991,
"venue": "Language and Speech",
"volume": "34",
"issue": "",
"pages": "351--366",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Anderson, A. H., Bader, M., Bard, E. G., Boyle, E., Do- herty, G., Garrod, S., Isard, S., Kowtko, J., McAllister, J., Miller, J., et al. (1991). The HCRC Map Task corpus. Language and Speech, 34:351-366.",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "Speaking while monitoring addressees for understanding",
"authors": [
{
"first": "H",
"middle": [
"H"
],
"last": "Clark",
"suffix": ""
},
{
"first": "M",
"middle": [
"A"
],
"last": "Krych",
"suffix": ""
}
],
"year": 2004,
"venue": "Journal of Memory and Language",
"volume": "50",
"issue": "1",
"pages": "62--81",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Clark, H. H. and Krych, M. A. (2004). Speaking while monitoring addressees for understanding. Journal of Memory and Language, 50(1):62-81.",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "Where's wally: the influence of visual salience on referring expression generation",
"authors": [
{
"first": "A",
"middle": [
"D"
],
"last": "Clarke",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Elsner",
"suffix": ""
},
{
"first": "H",
"middle": [],
"last": "Rohde",
"suffix": ""
}
],
"year": 2013,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Clarke, A. D., Elsner, M., and Rohde, H. (2013). Where's wally: the influence of visual salience on referring ex- pression generation. Frontiers in psychology, 4.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "Interaction in task-oriented human-human dialogue: The effects of different turn-taking policies. Proceedings of the First International IEEE",
"authors": [
{
"first": "R",
"middle": [],
"last": "Fern\u00e1ndez",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Lucht",
"suffix": ""
},
{
"first": "K",
"middle": [],
"last": "Rodr\u00edguez",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Schlangen",
"suffix": ""
}
],
"year": 2006,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Fern\u00e1ndez, R., Lucht, T., Rodr\u00edguez, K., and Schlangen, D. (2006). Interaction in task-oriented human-human dia- logue: The effects of different turn-taking policies. Pro- ceedings of the First International IEEE/ACL Workshop on Spoken Language Technology.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "Pushto-talk ain't always bad! comparing different interactivity settings in task-oriented dialogue",
"authors": [
{
"first": "R",
"middle": [],
"last": "Fern\u00e1ndez",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Schlangen",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Lucht",
"suffix": ""
}
],
"year": 2007,
"venue": "Proceeding of DECALOG, the 11th International Workshop on the Semantics and Pragmatics of Dialogue",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Fern\u00e1ndez, R., Schlangen, D., and Lucht, T. (2007). Push- to-talk ain't always bad! comparing different interac- tivity settings in task-oriented dialogue. Proceeding of DECALOG, the 11th International Workshop on the Se- mantics and Pragmatics of Dialogue (SemDial07).",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "Learning when to point: A data-driven approach",
"authors": [
{
"first": "A",
"middle": [],
"last": "Gatt",
"suffix": ""
},
{
"first": "P",
"middle": [],
"last": "Paggio",
"suffix": ""
}
],
"year": 2014,
"venue": "Proceedings of COLING 2014, the 25th International Conference on Computational Linguistics: Technical Papers",
"volume": "",
"issue": "",
"pages": "2007--2017",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Gatt, A. and Paggio, P. (2014). Learning when to point: A data-driven approach. In Proceedings of COLING 2014, the 25th International Conference on Computa- tional Linguistics: Technical Papers, pages 2007-2017, Dublin, Ireland, aug. Dublin City University and Associ- ation for Computational Linguistics.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "From the virtual to the real world: Referring to objects in real-world spatial scenes",
"authors": [
{
"first": "D",
"middle": [],
"last": "Gkatzia",
"suffix": ""
},
{
"first": "V",
"middle": [],
"last": "Rieser",
"suffix": ""
},
{
"first": "P",
"middle": [],
"last": "Bartie",
"suffix": ""
},
{
"first": "W",
"middle": [],
"last": "Mackaness",
"suffix": ""
}
],
"year": 2015,
"venue": "Proceedings of EMNLP 2015. Association for Computational Linguistics",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Gkatzia, D., Rieser, V., Bartie, P., and Mackaness, W. (2015). From the virtual to the real world: Referring to objects in real-world spatial scenes. In Proceedings of EMNLP 2015. Association for Computational Linguis- tics.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "Comparing local and sequential models for statistical incremental natural language understanding",
"authors": [
{
"first": "S",
"middle": [],
"last": "Heintze",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Baumann",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Schlangen",
"suffix": ""
}
],
"year": 2010,
"venue": "Proceedings of the 11th Annual Meeting of the Special Interest Group on Discourse and Dialogue",
"volume": "",
"issue": "",
"pages": "9--16",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Heintze, S., Baumann, T., and Schlangen, D. (2010). Com- paring local and sequential models for statistical incre- mental natural language understanding. In Proceedings of the 11th Annual Meeting of the Special Interest Group on Discourse and Dialogue, pages 9-16. Association for Computational Linguistics.",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "Disfluency and laughter annotation in a lightweight dialogue mark-up protocol",
"authors": [
{
"first": "J",
"middle": [],
"last": "Hough",
"suffix": ""
},
{
"first": "L",
"middle": [],
"last": "De Ruiter",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Betz",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Schlangen",
"suffix": ""
}
],
"year": 2015,
"venue": "The 6th Workshop on Disfluency in Spontaneous Speech (DiSS)",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Hough, J., de Ruiter, L., Betz, S., and Schlangen, D. (2015). Disfluency and laughter annotation in a light- weight dialogue mark-up protocol. In The 6th Workshop on Disfluency in Spontaneous Speech (DiSS).",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "ReferItGame: Referring to Objects in Photographs of Natural Scenes",
"authors": [
{
"first": "S",
"middle": [],
"last": "Kazemzadeh",
"suffix": ""
},
{
"first": "V",
"middle": [],
"last": "Ordonez",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Matten",
"suffix": ""
},
{
"first": "T",
"middle": [
"L"
],
"last": "Berg",
"suffix": ""
}
],
"year": 2014,
"venue": "Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP 2014)",
"volume": "",
"issue": "",
"pages": "787--798",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Kazemzadeh, S., Ordonez, V., Matten, M., and Berg, T. L. (2014). ReferItGame: Referring to Objects in Pho- tographs of Natural Scenes. In Proceedings of the Con- ference on Empirical Methods in Natural Language Pro- cessing (EMNLP 2014), pages 787-798, Doha, Qatar.",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "Simple learning and compositional application of perceptually grounded word meanings for incremental reference resolution",
"authors": [
{
"first": "C",
"middle": [],
"last": "Kennington",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Schlangen",
"suffix": ""
}
],
"year": 2015,
"venue": "Proceedings of the Conference for the Association for Computational Linguistics (ACL)",
"volume": "",
"issue": "",
"pages": "292--301",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Kennington, C. and Schlangen, D. (2015). Simple learning and compositional application of perceptually grounded word meanings for incremental reference resolution. Proceedings of the Conference for the Association for Computational Linguistics (ACL), pages 292-301. As- sociation for Computational Linguistics.",
"links": null
},
"BIBREF11": {
"ref_id": "b11",
"title": "Interpreting situated dialogue utterances: an update model that uses speech, gaze, and gesture information",
"authors": [
{
"first": "C",
"middle": [],
"last": "Kennington",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Kousidis",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Schlangen",
"suffix": ""
}
],
"year": 2013,
"venue": "Proceedings of SIGdial",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Kennington, C., Kousidis, S., and Schlangen, D. (2013). Interpreting situated dialogue utterances: an update model that uses speech, gaze, and gesture information. Proceedings of SIGdial 2013.",
"links": null
},
"BIBREF12": {
"ref_id": "b12",
"title": "The d-tuna corpus: A dutch dataset for the evaluation of referring expression generation algorithms",
"authors": [
{
"first": "R",
"middle": [],
"last": "Koolen",
"suffix": ""
},
{
"first": "E",
"middle": [],
"last": "Krahmer",
"suffix": ""
}
],
"year": 2010,
"venue": "LREC",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Koolen, R. and Krahmer, E. (2010). The d-tuna corpus: A dutch dataset for the evaluation of referring expression generation algorithms. In LREC.",
"links": null
},
"BIBREF13": {
"ref_id": "b13",
"title": "Evaluating a minimally invasive laboratory architecture for recording multimodal conversational data",
"authors": [
{
"first": "S",
"middle": [],
"last": "Kousidis",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Pfeiffer",
"suffix": ""
},
{
"first": "Z",
"middle": [],
"last": "Malisz",
"suffix": ""
},
{
"first": "P",
"middle": [],
"last": "Wagner",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Schlangen",
"suffix": ""
}
],
"year": 2012,
"venue": "Proceedings of the Interdisciplinary Workshop on Feedback Behaviors in Dialog, INTER-SPEECH2012 Satellite Workshop",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Kousidis, S., Pfeiffer, T., Malisz, Z., Wagner, P., and Schlangen, D. (2012). Evaluating a minimally invasive laboratory architecture for recording multimodal conver- sational data. In Proceedings of the Interdisciplinary Workshop on Feedback Behaviors in Dialog, INTER- SPEECH2012 Satellite Workshop.",
"links": null
},
"BIBREF14": {
"ref_id": "b14",
"title": "Walk the talk: Connecting language, knowledge, and action in route instructions",
"authors": [
{
"first": "M",
"middle": [],
"last": "Macmahon",
"suffix": ""
},
{
"first": "B",
"middle": [],
"last": "Stankiewicz",
"suffix": ""
},
{
"first": "B",
"middle": [],
"last": "Kuipers",
"suffix": ""
}
],
"year": 2006,
"venue": "Def",
"volume": "",
"issue": "",
"pages": "1475--1482",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "MacMahon, M., Stankiewicz, B., and Kuipers, B. (2006). Walk the talk: Connecting language, knowledge, and ac- tion in route instructions. Def, pages 1475-1482.",
"links": null
},
"BIBREF15": {
"ref_id": "b15",
"title": "Reducing the cost of dialogue system training and evaluation with online, crowd-sourced dialogue data collection",
"authors": [
{
"first": "R",
"middle": [],
"last": "Manuvinakurike",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Paetzel",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Devault",
"suffix": ""
}
],
"year": 2015,
"venue": "Proceedings of SEMDIAL 2015 goDIAL",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Manuvinakurike, R., Paetzel, M., and DeVault, D. (2015). Reducing the cost of dialogue system training and eval- uation with online, crowd-sourced dialogue data collec- tion. In Proceedings of SEMDIAL 2015 goDIAL.",
"links": null
},
"BIBREF16": {
"ref_id": "b16",
"title": "Natural reference to objects in a visual domain",
"authors": [
{
"first": "M",
"middle": [],
"last": "Mitchell",
"suffix": ""
},
{
"first": "K",
"middle": [],
"last": "Van Deemter",
"suffix": ""
},
{
"first": "E",
"middle": [],
"last": "Reiter",
"suffix": ""
}
],
"year": 2010,
"venue": "Proceedings of the 6th international natural language generation conference",
"volume": "",
"issue": "",
"pages": "95--104",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Mitchell, M., van Deemter, K., and Reiter, E. (2010). Nat- ural reference to objects in a visual domain. In Proceed- ings of the 6th international natural language genera- tion conference, pages 95-104. Association for Compu- tational Linguistics.",
"links": null
},
"BIBREF17": {
"ref_id": "b17",
"title": "Speaking through a noisy channel -experiments on inducing clarification behaviour in human-human dialogue",
"authors": [
{
"first": "D",
"middle": [],
"last": "Schlangen",
"suffix": ""
},
{
"first": "R",
"middle": [],
"last": "Fern\u00e1ndez",
"suffix": ""
}
],
"year": 2007,
"venue": "Proceedings of Interspeech",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Schlangen, D. and Fern\u00e1ndez, R. (2007). Speaking through a noisy channel -experiments on inducing clar- ification behaviour in human-human dialogue. Proceed- ings of Interspeech 2007.",
"links": null
},
"BIBREF18": {
"ref_id": "b18",
"title": "Incremental reference resolution: The task, metrics for evaluation, and a bayesian filtering model that is sensitive to disfluencies",
"authors": [
{
"first": "D",
"middle": [],
"last": "Schlangen",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Baumann",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Atterer",
"suffix": ""
}
],
"year": 2009,
"venue": "Proceedings of SIGdial 2009, the 10th Annual SIGDIAL Meeting on Discourse and Dialogue",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Schlangen, D., Baumann, T., and Atterer, M. (2009). In- cremental reference resolution: The task, metrics for evaluation, and a bayesian filtering model that is sensi- tive to disfluencies. Proceedings of SIGdial 2009, the 10th Annual SIGDIAL Meeting on Discourse and Dia- logue.",
"links": null
},
"BIBREF19": {
"ref_id": "b19",
"title": "The REX corpora : A collection of multimodal corpora of referring expressions in collaborative problem solving dialogues",
"authors": [
{
"first": "T",
"middle": [],
"last": "Tokunaga",
"suffix": ""
},
{
"first": "R",
"middle": [],
"last": "Iida",
"suffix": ""
},
{
"first": "A",
"middle": [],
"last": "Terai",
"suffix": ""
},
{
"first": "N",
"middle": [],
"last": "Kuriyama",
"suffix": ""
}
],
"year": 2012,
"venue": "Proceedings of the Eigth International Conference on Language Resources and Evaluation (LREC 2012)",
"volume": "",
"issue": "",
"pages": "422--429",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Tokunaga, T., Iida, R., Terai, A., and Kuriyama, N. (2012). The REX corpora : A collection of multimodal cor- pora of referring expressions in collaborative problem solving dialogues. In Proceedings of the Eigth Interna- tional Conference on Language Resources and Evalua- tion (LREC 2012), pages 422-429.",
"links": null
}
},
"ref_entries": {
"FIGREF1": {
"num": null,
"uris": null,
"text": "Database design for representing the mapping between dynamic visual context, words and references",
"type_str": "figure"
},
"TABREF1": {
"text": "Overview of experimental settings in the PentoRef corpus real scenes were more cluttered, and pieces can have various orientations.",
"num": null,
"content": "<table/>",
"html": null,
"type_str": "table"
},
"TABREF7": {
"text": "Corpus statistics and available annotations for PentoRef thors. For a subset of our corpora, disfluency and laughter annotation is also included in-line in the way described therein, making it suitable for training and testing disfluency detection. For a subset of the corpora the segments are given dialogue act type tags such as Instruction, Confirmation and ClarificationRequest.",
"num": null,
"content": "<table/>",
"html": null,
"type_str": "table"
}
}
}
} |