File size: 36,716 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 |
{
"paper_id": "W11-0139",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T05:38:29.793409Z"
},
"title": "BALLGAME: A Corpus for Computational Semantics",
"authors": [
{
"first": "Ezra",
"middle": [],
"last": "Keshet",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "University of Michigan",
"location": {}
},
"email": "[email protected]"
},
{
"first": "Terry",
"middle": [],
"last": "Szymanski",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "University of Michigan",
"location": {}
},
"email": ""
},
{
"first": "Stephen",
"middle": [],
"last": "Tyndall",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "University of Michigan",
"location": {}
},
"email": "[email protected]"
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "In this paper, we describe the Baseball Announcers' Language Linked with General Annotation of Meaningful Events (BALLGAME) project-a text corpus for research in computional semantics. We collected pitch-by-pitch event data for a sample of baseball games and used this data to build an annotated corpus composed of transcripts of radio broadcasts of these games. Our annotation links text from the broadcast to events in a formal representation of the semantics of the baseball game. We describe our corpus model, the annotation tool used to create the corpus, and conclude by discussing applications of this corpus in semantics research and natural language processing.",
"pdf_parse": {
"paper_id": "W11-0139",
"_pdf_hash": "",
"abstract": [
{
"text": "In this paper, we describe the Baseball Announcers' Language Linked with General Annotation of Meaningful Events (BALLGAME) project-a text corpus for research in computional semantics. We collected pitch-by-pitch event data for a sample of baseball games and used this data to build an annotated corpus composed of transcripts of radio broadcasts of these games. Our annotation links text from the broadcast to events in a formal representation of the semantics of the baseball game. We describe our corpus model, the annotation tool used to create the corpus, and conclude by discussing applications of this corpus in semantics research and natural language processing.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "The use of large annotated corpora and treebanks has led to many fruitful research programs in computational linguistics. At the time of this writing, Marcus et al. (1993) , which introduces the University of Pennsylvania Treebank, 1 has been cited by over 3000 subsequent papers. 2 Such treebanks are invaluable for the training and testing of large-scale syntactic parsers and numerous other applications in the field of Computational Syntax.",
"cite_spans": [
{
"start": 151,
"end": 171,
"text": "Marcus et al. (1993)",
"ref_id": "BIBREF5"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "Unfortunately for the field of Computational Semantics, there are few corresponding annotated corpora or treebanks representing the formalized meaning of natural language sentences, mainly because there is very little agreement on what such a representation of meaning would look like for arbitrary text. To overcome this obstacle, several recent studies have turned to the arena of sports, pairing natural language with game statistics in several domains, including RoboCup soccer (Liang et al., 2009; Chen et al., 2010) , soccer (Theune and Klabbers, 1998; Saggion et al., 2003) , American football (Barzilay and Lapata, 2005; Liang et al., 2009) , and baseball (Fleischman, 2007) .",
"cite_spans": [
{
"start": 482,
"end": 502,
"text": "(Liang et al., 2009;",
"ref_id": "BIBREF4"
},
{
"start": 503,
"end": 521,
"text": "Chen et al., 2010)",
"ref_id": "BIBREF1"
},
{
"start": 531,
"end": 558,
"text": "(Theune and Klabbers, 1998;",
"ref_id": "BIBREF7"
},
{
"start": 559,
"end": 580,
"text": "Saggion et al., 2003)",
"ref_id": "BIBREF6"
},
{
"start": 601,
"end": 628,
"text": "(Barzilay and Lapata, 2005;",
"ref_id": "BIBREF0"
},
{
"start": 629,
"end": 648,
"text": "Liang et al., 2009)",
"ref_id": "BIBREF4"
},
{
"start": 664,
"end": 682,
"text": "(Fleischman, 2007)",
"ref_id": "BIBREF3"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "We have adapted this approach in the creation of a semantics-oriented corpus, using the domain of major-league baseball. The information state of a baseball game can be represented with a small number of variables, such as who is on which base, who is batting, who is playing each position, and the current score and inning. There is even a standard way of representing updates to this information state. 3 This makes baseball a logical stepping stone to a fuller representation of the world. We also chose baseball for this corpus because of the volume of data available, in the form of both natural language descriptions of events and language-independent game statistics. Most of professional baseball's thousands of games per year have at least two television broadcasts (home and away) and at least two radio broadcasts, often in multiple languages. The scorecard statistics for each game are also kept and made available on the internet, along with complete ordered lists of in-game events. These resources, coupled with a highcoverage syntactic parser, allow one to link natural language utterances with representations of their syntax and semantics.",
"cite_spans": [
{
"start": 405,
"end": 406,
"text": "3",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "The basic design of the BALLGAME corpus is a mapping between spans of text and events in a baseball game. The raw text comes from the transcribed speech of announcers broadcasting the radio play-byplay of a professional baseball game. This text is chunked into spans, and these spans are then labeled according to the following scheme:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Corpus Design",
"sec_num": "2"
},
{
"text": "\u2022 Event is the label given to a span that describes an event in our representation of the game for the first time. (Examples of events are simultaneous descriptions of pitches, plays, and stolen bases.)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Corpus Design",
"sec_num": "2"
},
{
"text": "\u2022 Recap is the label given to a span that correlates with prior events in the game. (Examples of recaps are when the announcer states the current score or strike count, or summarizes the current batter's previous at-bats.)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Corpus Design",
"sec_num": "2"
},
{
"text": "\u2022 Banter is the label given to a span that does not relate to an event in the game. The majority of spans are labeled as banter. (Examples of banters are \"color\" commentary, any discussion of the day's news, other baseball games, advertisements, etc.)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Corpus Design",
"sec_num": "2"
},
{
"text": "The term \"span\" has no linguistic significance, although spans often turn out to be sentences or clauses. Each span from the text that is labeled as an event is linked to one or more events in the model of the game as shown in Figure 1 . Not every event is linked to a span of text, since some events go unmentioned by the announcers. We model each game as a time-ordered sequence of baseball events, designed so that the state of the game at any given point, while not explicitly represented, can be computed given the events from the start of the game up to that point. We use a simple event model inspired by the comprehensive scoring system developed by Retrosheet, 4 but modified to match our needs and data resources. For example, most baseball scoring systems are at-bat-based, but this system is too coarse-grained for our purposes. Therefore, we use a system in which the fundamental event type is the pitch. Every baseball action from the start of the pitcher's motion until the end of the play (a hit or an out) is categorized as a PITCH event. Several other event types exist to accommodate other plays (e.g. balks, pick-offs), non-play actions (e.g. coaching visits to the mound, rain delays), and procedural activities (e.g. ejections, player substitutions).",
"cite_spans": [],
"ref_spans": [
{
"start": 227,
"end": 235,
"text": "Figure 1",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "Corpus Design",
"sec_num": "2"
},
{
"text": "In addition to a category, each event has multiple attribute values. The possible attributes depend on the category. A PITCH event, for example, has attributes describing the type, speed, and location of the pitch as well as whether it results in a ball, strike, play, etc. If the result is a play, then there are additional attributes describing the fielders involved in the defensive play. On the other hand, a PICKOFF event has different attributes, describing which base the ball was thrown to, whether it resulted in an out, etc. In the future, we plan to add syntactic parse information for each span such as that generated using the Stanford Parser (De Marneffe et al., 2006) . Using an explicit syntactic representation, like the one illustrated in figure 2, it will be possible to label more detailed correlations between the text and the meaning. Even without explicit annotation, statistical learning methods could be used to infer, e.g., that the word \"curveball\" in the sentence in figure 2 correlates with the semantic attribute type='curve', or that the word \"CC\" correlates with a specific PLAYER entity. While the annotations in the corpus exist only at the sentence or phrase level, this type of further processing could push the annotation down to the word level, facilitating the study of lexical semantics and semantic transformations of syntactic structures.",
"cite_spans": [
{
"start": 656,
"end": 682,
"text": "(De Marneffe et al., 2006)",
"ref_id": "BIBREF2"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Corpus Design",
"sec_num": "2"
},
{
"text": "Student transcribers use a custom-created transcription and annotation tool, illustrated in Figure 3 , to add data to the corpus. They listen to and transcribe the radio broadcast, while simultaneously chunking the text into spans as described above. Each span is labeled banter, event, or recap, and, if the span describes an event, the student selects the corresponding event(s) from the event column.",
"cite_spans": [],
"ref_spans": [
{
"start": 92,
"end": 100,
"text": "Figure 3",
"ref_id": "FIGREF2"
}
],
"eq_spans": [],
"section": "Corpus Creation",
"sec_num": "3"
},
{
"text": "Annotators have access to a style guide to encourage consistency. This guide sets out two main principles: first, the transcript of an inning, taken as a whole, should be read like a well-edited, consistently formatted document; and second, all and only the events explicitly mentioned by the radio announcers should be linked to events in the game model.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Corpus Creation",
"sec_num": "3"
},
{
"text": "Although spans are displayed as separate lines in the transcription tool, in order to maintain this first style principle, we ask the students to imagine that all spans of the transcript are pasted together in sequence to form a normal transcript of the game. Thus, they are asked not to put ellipses or dashes at the end of spans nor to capitalize the beginnings of spans that do not begin sentences. Also included in this principle is a standardized formatting style for baseball statistics, such as strike counts, scores, and batting averages, so that, for instance, \"the count is two and oh\" is transcribed \"the count is 2-0\".",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Corpus Creation",
"sec_num": "3"
},
{
"text": "The second principle set out in the annotation style guide is meant to ensure that the events linked to a particular utterance are as close as possible to the \"meaning\" of that utterance. Integral to this process is consistently distinguishing the categories of event, recap and banter. Since recap and banter spans do not relate to events in the model, it is important to keep them separate from the event spans to get the most accurate data. Even given the descriptions of these categories from section 2, ambiguous cases still do arise on occasion. For instance, one common difficulty is distinguishing event from recap when an announcer discusses a play immediately after it happens. In such cases, in keeping with our annotation principle, we use the rule of thumb that only new information is annotated as event; old information is recap. We also adopt the rule that only game events that are explicitly stated by the announcer should be linked to spans; for example, if the announcer merely states the name of the batter (e.g. \"Cust takes a first-pitch strike\") in the process of describing the first pitch of his at-bat, then this should not reference the ATBAT event that indicates the arrival of a new batter at the plate. On the other hand, an explicit mention (e.g. \"Here's Cust.\") should.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Corpus Creation",
"sec_num": "3"
},
{
"text": "In the final steps of the annotation process, each transcript is reviewed and corrected by a second annotator to reduce errors and further promote consistency across annotators. ",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Corpus Creation",
"sec_num": "3"
},
{
"text": "Since this corpus links natural language utterances with complete semantic representations which fully describe the state of the baseball game, it has a number of applications for research in computational semantics. While the domain is limited, and the \"meaning\" of a baseball game does not approach the complexity of the possible \"meanings\" in the real world, nevertheless this corpus should be a useful resource both for developing NLP tools and for studying theories of language and meaning.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Potential Applications",
"sec_num": "4"
},
{
"text": "One application domain for this type of data is natural language generation and understanding, and much prior work connecting sports commentaries to statistics or events falls into this domain. One related generation task is to generate textual summaries of complete games: Theune and Klabbers (1998) generated spoken Dutch summaries of soccer matches, and Barzilay and Lapata (2005) investigate the relationship between textual NFL recaps and the box scores of the games. More similar to our project is the RoboCup announcer system of Chen et al. (2010) , which produces play-by-play commentary (in English and Korean) of simulated RoboCup soccer matches. Our corpus could certainly be used to train systems that predict the event structure given the text of the commentary, or vice-versa.",
"cite_spans": [
{
"start": 274,
"end": 300,
"text": "Theune and Klabbers (1998)",
"ref_id": "BIBREF7"
},
{
"start": 357,
"end": 383,
"text": "Barzilay and Lapata (2005)",
"ref_id": "BIBREF0"
},
{
"start": 536,
"end": 554,
"text": "Chen et al. (2010)",
"ref_id": "BIBREF1"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Potential Applications",
"sec_num": "4"
},
{
"text": "In the domain of information extraction, our corpus could be used to train systems to infer representations of meaning from texts. In many domains, the same word or phrase can appear in a variety of different contexts with different ramifications. For example, the phrase \"home run\" in a baseball commentary may mean that a home run has just occurred, or it may refer to a home run in a previous game, or a player's home-run totals for the season, etc.. Fleischman (2007) , using a collection of video broadcasts of baseball games, combines natural language processing with artificial vision technology to resolve when events like home runs actually occur, in order to facilitate retrieval of relevant video clips. Using our corpus, one could design a system to perform the same task based purely on the textual data, perhaps to extend this same task to radio broadcasts as well as television broadcasts. Given the corpus labels of event, recap, and banter, a classifier could be built to identify only the event regions, and an extraction system could identify the relevant semantic features (e.g. player names, types of events).",
"cite_spans": [
{
"start": 454,
"end": 471,
"text": "Fleischman (2007)",
"ref_id": "BIBREF3"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Potential Applications",
"sec_num": "4"
},
{
"text": "While generation and understanding are tasks most applicable to this corpus, we hope researchers will find additional innovative uses of the corpus. For example, given that we plan to incorporate a number of baseball games with commentary both in English and Spanish, there is a potential connection to machine translation, particularly approaches that utilize comparable (rather than parallel) corpora. In our corpus, the comparable sections (i.e. the event-labeled regions) are explicitly aligned with one another, which is not usually the case in comparable corpora. Also, the corpus could prove useful for research on formal semantics, despite the fact the meaning representation is not particularly rich compared to modern semantic theory, and the jargon and speech styles are very specific to the domain of baseball sportscasts.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Potential Applications",
"sec_num": "4"
},
{
"text": "We have presented an overview of the BALLGAME annotated corpus for research in computational semantics, as well as a description of our procedure for annotation and the specialized annotation tool we developed for this purpose. To date, the corpus contains sixteen three-to four-hour-long major league baseball radio broadcasts, transcribed and annotated as described above. This represents 237,100 transcribed words in 13,382 spans (6,511 banter; 3,994 event; 2,877 recap). Work is ongoing, and the goal is to complete fifty games by the end of the year. We believe this corpus, by pairing natural language text with formalized representations of meaning, will prove useful for many types of NLP research.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "5"
},
{
"text": "http://www.cis.upenn.edu/\u02dctreebank/ 2 http://scholar.google.com/scholar?cites=7124559111460341353 3 See example scorecards at http://swingleydev.com/baseball/tutorial.php.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "http://www.retrosheet.org",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
}
],
"back_matter": [],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "Collective content selection for concept-to-text generation",
"authors": [
{
"first": "R",
"middle": [],
"last": "Barzilay",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Lapata",
"suffix": ""
}
],
"year": 2005,
"venue": "Proceedings of HLT/EMNLP",
"volume": "",
"issue": "",
"pages": "331--338",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Barzilay, R. and M. Lapata (2005). Collective content selection for concept-to-text generation. In Pro- ceedings of HLT/EMNLP, pp. 331-338.",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "Training a multilingual sportscaster: Using perceptual context to learn language",
"authors": [
{
"first": "D",
"middle": [],
"last": "Chen",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Kim",
"suffix": ""
},
{
"first": "R",
"middle": [],
"last": "Mooney",
"suffix": ""
}
],
"year": 2010,
"venue": "Journal of Artificial Intelligence Research",
"volume": "37",
"issue": "1",
"pages": "397--436",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Chen, D., J. Kim, and R. Mooney (2010). Training a multilingual sportscaster: Using perceptual context to learn language. Journal of Artificial Intelligence Research 37(1), 397-436.",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "Generating typed dependency parses from phrase structure parses",
"authors": [
{
"first": "M",
"middle": [],
"last": "De Marneffe",
"suffix": ""
},
{
"first": "B",
"middle": [],
"last": "Maccartney",
"suffix": ""
},
{
"first": "C",
"middle": [],
"last": "Manning",
"suffix": ""
}
],
"year": 2006,
"venue": "LREC",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "De Marneffe, M., B. MacCartney, and C. Manning (2006). Generating typed dependency parses from phrase structure parses. In LREC 2006.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "Situated models of meaning for sports video retrieval",
"authors": [
{
"first": "M",
"middle": [],
"last": "Fleischman",
"suffix": ""
}
],
"year": 2007,
"venue": "NAACL-HLT 2007",
"volume": "",
"issue": "",
"pages": "37--40",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Fleischman, M. (2007). Situated models of meaning for sports video retrieval. In NAACL-HLT 2007, pp. 37-40.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "Learning semantic correspondences with less supervision",
"authors": [
{
"first": "P",
"middle": [],
"last": "Liang",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Jordan",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Klein",
"suffix": ""
}
],
"year": 2009,
"venue": "Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural Language Processing of the AFNLP",
"volume": "1",
"issue": "",
"pages": "91--99",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Liang, P., M. Jordan, and D. Klein (2009). Learning semantic correspondences with less supervision. In Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural Language Processing of the AFNLP: Volume 1, pp. 91-99.",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "Building a large annotated corpus of English: The Penn Treebank",
"authors": [
{
"first": "M",
"middle": [],
"last": "Marcus",
"suffix": ""
},
{
"first": "B",
"middle": [],
"last": "Santorini",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Marcinkiewicz",
"suffix": ""
}
],
"year": 1993,
"venue": "Computational linguistics",
"volume": "19",
"issue": "2",
"pages": "313--330",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Marcus, M., B. Santorini, and M. Marcinkiewicz (1993). Building a large annotated corpus of English: The Penn Treebank. Computational linguistics 19(2), 313-330.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "Event-coreference across multiple, multi-lingual sources in the Mumis project",
"authors": [
{
"first": "H",
"middle": [],
"last": "Saggion",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Kuper",
"suffix": ""
},
{
"first": "H",
"middle": [],
"last": "Cunningham",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Declerck",
"suffix": ""
},
{
"first": "P",
"middle": [],
"last": "Wittenburg",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Puts",
"suffix": ""
},
{
"first": "E",
"middle": [],
"last": "Hoenkamp",
"suffix": ""
},
{
"first": "F",
"middle": [],
"last": "Jong",
"suffix": ""
},
{
"first": "Y",
"middle": [],
"last": "Wilks",
"suffix": ""
}
],
"year": 2003,
"venue": "Proceedings of the tenth conference on European chapter of the Association for Computational Linguistics",
"volume": "2",
"issue": "",
"pages": "239--242",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Saggion, H., J. Kuper, H. Cunningham, T. Declerck, P. Wittenburg, M. Puts, E. Hoenkamp, F. de Jong, and Y. Wilks (2003). Event-coreference across multiple, multi-lingual sources in the Mumis project. In Proceedings of the tenth conference on European chapter of the Association for Computational Linguistics: Volume 2, pp. 239-242.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "GoalGetter: Generation of spoken soccer reports",
"authors": [
{
"first": "M",
"middle": [],
"last": "Theune",
"suffix": ""
},
{
"first": "E",
"middle": [],
"last": "Klabbers",
"suffix": ""
}
],
"year": 1998,
"venue": "Proceedings of the Ninth International Workshop on Natural Language Generation",
"volume": "",
"issue": "",
"pages": "292--295",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Theune, M. and E. Klabbers (1998). GoalGetter: Generation of spoken soccer reports. In Proceedings of the Ninth International Workshop on Natural Language Generation, pp. 292-295.",
"links": null
}
},
"ref_entries": {
"FIGREF0": {
"text": "Illustration of a portion of the corpus: event spans of the text (on the left) are associated with events from a standardized description of the ballgame (on the right).",
"type_str": "figure",
"num": null,
"uris": null
},
"FIGREF1": {
"text": "Example of a dependency parsed transcript line and corresponding events.",
"type_str": "figure",
"num": null,
"uris": null
},
"FIGREF2": {
"text": "Screen shot of online annotation tool.",
"type_str": "figure",
"num": null,
"uris": null
}
}
}
} |