|
a |
|
b/test-resources/doc-features.json |
|
|
1 |
{ |
|
|
2 |
"text": "He loved to smoke but Marlboro cigarettes gave John Smith lung cancer while he was in Chicago.", |
|
|
3 |
"sentences": [ |
|
|
4 |
{ |
|
|
5 |
"text": "He loved to smoke but Marlboro cigarettes gave John Smith lung cancer while he was in Chicago.", |
|
|
6 |
"tokens": [ |
|
|
7 |
{ |
|
|
8 |
"sent_i": 0, |
|
|
9 |
"norm_len": 2, |
|
|
10 |
"i_sent": 0, |
|
|
11 |
"lexspan": { |
|
|
12 |
"end": 2, |
|
|
13 |
"begin": 0 |
|
|
14 |
}, |
|
|
15 |
"is_wh": false, |
|
|
16 |
"i": 0, |
|
|
17 |
"is_superlative": false, |
|
|
18 |
"shape": 12204527652707022206, |
|
|
19 |
"is_stop": true, |
|
|
20 |
"dep": 429, |
|
|
21 |
"ent": 0, |
|
|
22 |
"norm": "He", |
|
|
23 |
"pos_": "PRON", |
|
|
24 |
"lemma_": "he", |
|
|
25 |
"idx": 0, |
|
|
26 |
"tag": 13656873538139661788, |
|
|
27 |
"dep_": "nsubj", |
|
|
28 |
"is_punctuation": false, |
|
|
29 |
"ent_iob": 0, |
|
|
30 |
"tag_": "PRP", |
|
|
31 |
"is_ent": false, |
|
|
32 |
"ent_iob_": "O", |
|
|
33 |
"children": [], |
|
|
34 |
"is_space": false, |
|
|
35 |
"ent_": "-<N>-", |
|
|
36 |
"shape_": "Xx", |
|
|
37 |
"is_pronoun": true, |
|
|
38 |
"is_contraction": false, |
|
|
39 |
"cui": -1, |
|
|
40 |
"definition_": "-<N>-", |
|
|
41 |
"tuis": "frozenset()", |
|
|
42 |
"detected_name_": "-<N>-", |
|
|
43 |
"sub_names": [], |
|
|
44 |
"cui_": "-<N>-", |
|
|
45 |
"tui_descs_": "", |
|
|
46 |
"pref_name_": "-<N>-", |
|
|
47 |
"is_concept": false, |
|
|
48 |
"tuis_": "" |
|
|
49 |
}, |
|
|
50 |
{ |
|
|
51 |
"sent_i": 0, |
|
|
52 |
"norm_len": 5, |
|
|
53 |
"i_sent": 1, |
|
|
54 |
"lexspan": { |
|
|
55 |
"end": 8, |
|
|
56 |
"begin": 3 |
|
|
57 |
}, |
|
|
58 |
"is_wh": false, |
|
|
59 |
"i": 1, |
|
|
60 |
"is_superlative": false, |
|
|
61 |
"shape": 13110060611322374290, |
|
|
62 |
"is_stop": false, |
|
|
63 |
"dep": 8206900633647566924, |
|
|
64 |
"ent": 0, |
|
|
65 |
"norm": "loved", |
|
|
66 |
"pos_": "VERB", |
|
|
67 |
"lemma_": "love", |
|
|
68 |
"idx": 3, |
|
|
69 |
"tag": 17109001835818727656, |
|
|
70 |
"dep_": "ROOT", |
|
|
71 |
"is_punctuation": false, |
|
|
72 |
"ent_iob": 0, |
|
|
73 |
"tag_": "VBD", |
|
|
74 |
"is_ent": false, |
|
|
75 |
"ent_iob_": "O", |
|
|
76 |
"children": [ |
|
|
77 |
0, |
|
|
78 |
3, |
|
|
79 |
4, |
|
|
80 |
7 |
|
|
81 |
], |
|
|
82 |
"is_space": false, |
|
|
83 |
"ent_": "-<N>-", |
|
|
84 |
"shape_": "xxxx", |
|
|
85 |
"is_pronoun": false, |
|
|
86 |
"is_contraction": false, |
|
|
87 |
"cui": -1, |
|
|
88 |
"definition_": "-<N>-", |
|
|
89 |
"tuis": "frozenset()", |
|
|
90 |
"detected_name_": "-<N>-", |
|
|
91 |
"sub_names": [], |
|
|
92 |
"cui_": "-<N>-", |
|
|
93 |
"tui_descs_": "", |
|
|
94 |
"pref_name_": "-<N>-", |
|
|
95 |
"is_concept": false, |
|
|
96 |
"tuis_": "" |
|
|
97 |
}, |
|
|
98 |
{ |
|
|
99 |
"sent_i": 0, |
|
|
100 |
"norm_len": 2, |
|
|
101 |
"i_sent": 2, |
|
|
102 |
"lexspan": { |
|
|
103 |
"end": 11, |
|
|
104 |
"begin": 9 |
|
|
105 |
}, |
|
|
106 |
"is_wh": false, |
|
|
107 |
"i": 2, |
|
|
108 |
"is_superlative": false, |
|
|
109 |
"shape": 4370460163704169311, |
|
|
110 |
"is_stop": true, |
|
|
111 |
"dep": 405, |
|
|
112 |
"ent": 0, |
|
|
113 |
"norm": "to", |
|
|
114 |
"pos_": "PART", |
|
|
115 |
"lemma_": "to", |
|
|
116 |
"idx": 9, |
|
|
117 |
"tag": 5595707737748328492, |
|
|
118 |
"dep_": "aux", |
|
|
119 |
"is_punctuation": false, |
|
|
120 |
"ent_iob": 0, |
|
|
121 |
"tag_": "TO", |
|
|
122 |
"is_ent": false, |
|
|
123 |
"ent_iob_": "O", |
|
|
124 |
"children": [], |
|
|
125 |
"is_space": false, |
|
|
126 |
"ent_": "-<N>-", |
|
|
127 |
"shape_": "xx", |
|
|
128 |
"is_pronoun": false, |
|
|
129 |
"is_contraction": false, |
|
|
130 |
"cui": -1, |
|
|
131 |
"definition_": "-<N>-", |
|
|
132 |
"tuis": "frozenset()", |
|
|
133 |
"detected_name_": "-<N>-", |
|
|
134 |
"sub_names": [], |
|
|
135 |
"cui_": "-<N>-", |
|
|
136 |
"tui_descs_": "", |
|
|
137 |
"pref_name_": "-<N>-", |
|
|
138 |
"is_concept": false, |
|
|
139 |
"tuis_": "" |
|
|
140 |
}, |
|
|
141 |
{ |
|
|
142 |
"sent_i": 0, |
|
|
143 |
"norm_len": 5, |
|
|
144 |
"i_sent": 3, |
|
|
145 |
"lexspan": { |
|
|
146 |
"end": 17, |
|
|
147 |
"begin": 12 |
|
|
148 |
}, |
|
|
149 |
"is_wh": false, |
|
|
150 |
"i": 3, |
|
|
151 |
"is_superlative": false, |
|
|
152 |
"shape": 13110060611322374290, |
|
|
153 |
"is_stop": false, |
|
|
154 |
"dep": 450, |
|
|
155 |
"ent": 0, |
|
|
156 |
"norm": "smoke", |
|
|
157 |
"pos_": "VERB", |
|
|
158 |
"lemma_": "smoke", |
|
|
159 |
"idx": 12, |
|
|
160 |
"tag": 14200088355797579614, |
|
|
161 |
"dep_": "xcomp", |
|
|
162 |
"is_punctuation": false, |
|
|
163 |
"ent_iob": 0, |
|
|
164 |
"tag_": "VB", |
|
|
165 |
"is_ent": false, |
|
|
166 |
"ent_iob_": "O", |
|
|
167 |
"children": [ |
|
|
168 |
2 |
|
|
169 |
], |
|
|
170 |
"is_space": false, |
|
|
171 |
"ent_": "-<N>-", |
|
|
172 |
"shape_": "xxxx", |
|
|
173 |
"is_pronoun": false, |
|
|
174 |
"is_contraction": false, |
|
|
175 |
"cui": -1, |
|
|
176 |
"definition_": "-<N>-", |
|
|
177 |
"tuis": "frozenset()", |
|
|
178 |
"detected_name_": "-<N>-", |
|
|
179 |
"sub_names": [], |
|
|
180 |
"cui_": "-<N>-", |
|
|
181 |
"tui_descs_": "", |
|
|
182 |
"pref_name_": "-<N>-", |
|
|
183 |
"is_concept": false, |
|
|
184 |
"tuis_": "" |
|
|
185 |
}, |
|
|
186 |
{ |
|
|
187 |
"sent_i": 0, |
|
|
188 |
"norm_len": 3, |
|
|
189 |
"i_sent": 4, |
|
|
190 |
"lexspan": { |
|
|
191 |
"end": 21, |
|
|
192 |
"begin": 18 |
|
|
193 |
}, |
|
|
194 |
"is_wh": false, |
|
|
195 |
"i": 4, |
|
|
196 |
"is_superlative": false, |
|
|
197 |
"shape": 4088098365541558500, |
|
|
198 |
"is_stop": true, |
|
|
199 |
"dep": 407, |
|
|
200 |
"ent": 0, |
|
|
201 |
"norm": "but", |
|
|
202 |
"pos_": "CCONJ", |
|
|
203 |
"lemma_": "but", |
|
|
204 |
"idx": 18, |
|
|
205 |
"tag": 17571114184892886314, |
|
|
206 |
"dep_": "cc", |
|
|
207 |
"is_punctuation": false, |
|
|
208 |
"ent_iob": 0, |
|
|
209 |
"tag_": "CC", |
|
|
210 |
"is_ent": false, |
|
|
211 |
"ent_iob_": "O", |
|
|
212 |
"children": [], |
|
|
213 |
"is_space": false, |
|
|
214 |
"ent_": "-<N>-", |
|
|
215 |
"shape_": "xxx", |
|
|
216 |
"is_pronoun": false, |
|
|
217 |
"is_contraction": false, |
|
|
218 |
"cui": -1, |
|
|
219 |
"definition_": "-<N>-", |
|
|
220 |
"tuis": "frozenset()", |
|
|
221 |
"detected_name_": "-<N>-", |
|
|
222 |
"sub_names": [], |
|
|
223 |
"cui_": "-<N>-", |
|
|
224 |
"tui_descs_": "", |
|
|
225 |
"pref_name_": "-<N>-", |
|
|
226 |
"is_concept": false, |
|
|
227 |
"tuis_": "" |
|
|
228 |
}, |
|
|
229 |
{ |
|
|
230 |
"sent_i": 0, |
|
|
231 |
"norm_len": 8, |
|
|
232 |
"i_sent": 5, |
|
|
233 |
"lexspan": { |
|
|
234 |
"end": 30, |
|
|
235 |
"begin": 22 |
|
|
236 |
}, |
|
|
237 |
"is_wh": false, |
|
|
238 |
"i": 5, |
|
|
239 |
"is_superlative": false, |
|
|
240 |
"shape": 16072095006890171862, |
|
|
241 |
"is_stop": false, |
|
|
242 |
"dep": 7037928807040764755, |
|
|
243 |
"ent": 383, |
|
|
244 |
"norm": "Marlboro", |
|
|
245 |
"pos_": "PROPN", |
|
|
246 |
"lemma_": "Marlboro", |
|
|
247 |
"idx": 22, |
|
|
248 |
"tag": 15794550382381185553, |
|
|
249 |
"dep_": "compound", |
|
|
250 |
"is_punctuation": false, |
|
|
251 |
"ent_iob": 3, |
|
|
252 |
"tag_": "NNP", |
|
|
253 |
"is_ent": true, |
|
|
254 |
"ent_iob_": "B", |
|
|
255 |
"children": [], |
|
|
256 |
"is_space": false, |
|
|
257 |
"ent_": "ORG", |
|
|
258 |
"shape_": "Xxxxx", |
|
|
259 |
"is_pronoun": false, |
|
|
260 |
"is_contraction": false, |
|
|
261 |
"cui": -1, |
|
|
262 |
"definition_": "-<N>-", |
|
|
263 |
"tuis": "frozenset()", |
|
|
264 |
"detected_name_": "-<N>-", |
|
|
265 |
"sub_names": [], |
|
|
266 |
"cui_": "-<N>-", |
|
|
267 |
"tui_descs_": "", |
|
|
268 |
"pref_name_": "-<N>-", |
|
|
269 |
"is_concept": false, |
|
|
270 |
"tuis_": "" |
|
|
271 |
}, |
|
|
272 |
{ |
|
|
273 |
"sent_i": 0, |
|
|
274 |
"norm_len": 10, |
|
|
275 |
"i_sent": 6, |
|
|
276 |
"lexspan": { |
|
|
277 |
"end": 41, |
|
|
278 |
"begin": 31 |
|
|
279 |
}, |
|
|
280 |
"is_wh": false, |
|
|
281 |
"i": 6, |
|
|
282 |
"is_superlative": false, |
|
|
283 |
"shape": 13110060611322374290, |
|
|
284 |
"is_stop": false, |
|
|
285 |
"dep": 429, |
|
|
286 |
"ent": 0, |
|
|
287 |
"norm": "cigarettes", |
|
|
288 |
"pos_": "NOUN", |
|
|
289 |
"lemma_": "cigarette", |
|
|
290 |
"idx": 31, |
|
|
291 |
"tag": 783433942507015291, |
|
|
292 |
"dep_": "nsubj", |
|
|
293 |
"is_punctuation": false, |
|
|
294 |
"ent_iob": 0, |
|
|
295 |
"tag_": "NNS", |
|
|
296 |
"is_ent": false, |
|
|
297 |
"ent_iob_": "O", |
|
|
298 |
"children": [ |
|
|
299 |
5 |
|
|
300 |
], |
|
|
301 |
"is_space": false, |
|
|
302 |
"ent_": "-<N>-", |
|
|
303 |
"shape_": "xxxx", |
|
|
304 |
"is_pronoun": false, |
|
|
305 |
"is_contraction": false, |
|
|
306 |
"cui": -1, |
|
|
307 |
"definition_": "-<N>-", |
|
|
308 |
"tuis": "frozenset()", |
|
|
309 |
"detected_name_": "-<N>-", |
|
|
310 |
"sub_names": [], |
|
|
311 |
"cui_": "-<N>-", |
|
|
312 |
"tui_descs_": "", |
|
|
313 |
"pref_name_": "-<N>-", |
|
|
314 |
"is_concept": false, |
|
|
315 |
"tuis_": "" |
|
|
316 |
}, |
|
|
317 |
{ |
|
|
318 |
"sent_i": 0, |
|
|
319 |
"norm_len": 4, |
|
|
320 |
"i_sent": 7, |
|
|
321 |
"lexspan": { |
|
|
322 |
"end": 46, |
|
|
323 |
"begin": 42 |
|
|
324 |
}, |
|
|
325 |
"is_wh": false, |
|
|
326 |
"i": 7, |
|
|
327 |
"is_superlative": false, |
|
|
328 |
"shape": 13110060611322374290, |
|
|
329 |
"is_stop": false, |
|
|
330 |
"dep": 410, |
|
|
331 |
"ent": 0, |
|
|
332 |
"norm": "gave", |
|
|
333 |
"pos_": "VERB", |
|
|
334 |
"lemma_": "give", |
|
|
335 |
"idx": 42, |
|
|
336 |
"tag": 17109001835818727656, |
|
|
337 |
"dep_": "conj", |
|
|
338 |
"is_punctuation": false, |
|
|
339 |
"ent_iob": 0, |
|
|
340 |
"tag_": "VBD", |
|
|
341 |
"is_ent": false, |
|
|
342 |
"ent_iob_": "O", |
|
|
343 |
"children": [ |
|
|
344 |
6, |
|
|
345 |
11, |
|
|
346 |
14, |
|
|
347 |
17 |
|
|
348 |
], |
|
|
349 |
"is_space": false, |
|
|
350 |
"ent_": "-<N>-", |
|
|
351 |
"shape_": "xxxx", |
|
|
352 |
"is_pronoun": false, |
|
|
353 |
"is_contraction": false, |
|
|
354 |
"cui": -1, |
|
|
355 |
"definition_": "-<N>-", |
|
|
356 |
"tuis": "frozenset()", |
|
|
357 |
"detected_name_": "-<N>-", |
|
|
358 |
"sub_names": [], |
|
|
359 |
"cui_": "-<N>-", |
|
|
360 |
"tui_descs_": "", |
|
|
361 |
"pref_name_": "-<N>-", |
|
|
362 |
"is_concept": false, |
|
|
363 |
"tuis_": "" |
|
|
364 |
}, |
|
|
365 |
{ |
|
|
366 |
"sent_i": 0, |
|
|
367 |
"norm_len": 4, |
|
|
368 |
"i_sent": 8, |
|
|
369 |
"lexspan": { |
|
|
370 |
"end": 51, |
|
|
371 |
"begin": 47 |
|
|
372 |
}, |
|
|
373 |
"is_wh": false, |
|
|
374 |
"i": 8, |
|
|
375 |
"is_superlative": false, |
|
|
376 |
"shape": 10887629174180191697, |
|
|
377 |
"is_stop": false, |
|
|
378 |
"dep": 7037928807040764755, |
|
|
379 |
"ent": 380, |
|
|
380 |
"norm": "John", |
|
|
381 |
"pos_": "PROPN", |
|
|
382 |
"lemma_": "John", |
|
|
383 |
"idx": 47, |
|
|
384 |
"tag": 15794550382381185553, |
|
|
385 |
"dep_": "compound", |
|
|
386 |
"is_punctuation": false, |
|
|
387 |
"ent_iob": 3, |
|
|
388 |
"tag_": "NNP", |
|
|
389 |
"is_ent": true, |
|
|
390 |
"ent_iob_": "B", |
|
|
391 |
"children": [], |
|
|
392 |
"is_space": false, |
|
|
393 |
"ent_": "PERSON", |
|
|
394 |
"shape_": "Xxxx", |
|
|
395 |
"is_pronoun": false, |
|
|
396 |
"is_contraction": false, |
|
|
397 |
"cui": -1, |
|
|
398 |
"definition_": "-<N>-", |
|
|
399 |
"tuis": "frozenset()", |
|
|
400 |
"detected_name_": "-<N>-", |
|
|
401 |
"sub_names": [], |
|
|
402 |
"cui_": "-<N>-", |
|
|
403 |
"tui_descs_": "", |
|
|
404 |
"pref_name_": "-<N>-", |
|
|
405 |
"is_concept": false, |
|
|
406 |
"tuis_": "" |
|
|
407 |
}, |
|
|
408 |
{ |
|
|
409 |
"sent_i": 0, |
|
|
410 |
"norm_len": 5, |
|
|
411 |
"i_sent": 9, |
|
|
412 |
"lexspan": { |
|
|
413 |
"end": 57, |
|
|
414 |
"begin": 52 |
|
|
415 |
}, |
|
|
416 |
"is_wh": false, |
|
|
417 |
"i": 9, |
|
|
418 |
"is_superlative": false, |
|
|
419 |
"shape": 16072095006890171862, |
|
|
420 |
"is_stop": false, |
|
|
421 |
"dep": 7037928807040764755, |
|
|
422 |
"ent": 380, |
|
|
423 |
"norm": "Smith", |
|
|
424 |
"pos_": "PROPN", |
|
|
425 |
"lemma_": "Smith", |
|
|
426 |
"idx": 52, |
|
|
427 |
"tag": 15794550382381185553, |
|
|
428 |
"dep_": "compound", |
|
|
429 |
"is_punctuation": false, |
|
|
430 |
"ent_iob": 1, |
|
|
431 |
"tag_": "NNP", |
|
|
432 |
"is_ent": true, |
|
|
433 |
"ent_iob_": "I", |
|
|
434 |
"children": [ |
|
|
435 |
8 |
|
|
436 |
], |
|
|
437 |
"is_space": false, |
|
|
438 |
"ent_": "PERSON", |
|
|
439 |
"shape_": "Xxxxx", |
|
|
440 |
"is_pronoun": false, |
|
|
441 |
"is_contraction": false, |
|
|
442 |
"cui": 86418, |
|
|
443 |
"definition_": "-<N>-", |
|
|
444 |
"tuis": [ |
|
|
445 |
"T016" |
|
|
446 |
], |
|
|
447 |
"detected_name_": "smith", |
|
|
448 |
"sub_names": [ |
|
|
449 |
"allen~james~wilcox", |
|
|
450 |
"allen~j~.~wilcox", |
|
|
451 |
"amar~angela~f", |
|
|
452 |
"baity", |
|
|
453 |
"bandoh", |
|
|
454 |
"bostroem", |
|
|
455 |
"carlos~arvelo", |
|
|
456 |
"cotard", |
|
|
457 |
"danlos", |
|
|
458 |
"dr~parkes~weber", |
|
|
459 |
"ehlers", |
|
|
460 |
"foucauldian", |
|
|
461 |
"francisco~herrera~luque", |
|
|
462 |
"george~bernard~shaw", |
|
|
463 |
"giuseppe~paravicini", |
|
|
464 |
"hahn", |
|
|
465 |
"harris", |
|
|
466 |
"hilsenroth", |
|
|
467 |
"homo~sapien", |
|
|
468 |
"homo~sapiens", |
|
|
469 |
"human", |
|
|
470 |
"humanized~version", |
|
|
471 |
"humans", |
|
|
472 |
"human~being", |
|
|
473 |
"human~beings", |
|
|
474 |
"human~origin", |
|
|
475 |
"human~population", |
|
|
476 |
"jose~luis~vethencourt", |
|
|
477 |
"jose~solane", |
|
|
478 |
"jose~solanes", |
|
|
479 |
"jules~cotard", |
|
|
480 |
"knowles", |
|
|
481 |
"lisandro~alvarado", |
|
|
482 |
"max~clara", |
|
|
483 |
"mmy", |
|
|
484 |
"mmy~s", |
|
|
485 |
"parkes~weber", |
|
|
486 |
"pinel~s", |
|
|
487 |
"schwartz", |
|
|
488 |
"sekula~l~kathleen", |
|
|
489 |
"seligman~and~maier", |
|
|
490 |
"shimokawa", |
|
|
491 |
"single~human~host", |
|
|
492 |
"single~human~hosts", |
|
|
493 |
"smith", |
|
|
494 |
"stiles", |
|
|
495 |
"tchernabogov", |
|
|
496 |
"trump", |
|
|
497 |
"trump~s" |
|
|
498 |
], |
|
|
499 |
"cui_": "C0086418", |
|
|
500 |
"tui_descs_": "Human", |
|
|
501 |
"pref_name_": "Homo sapiens", |
|
|
502 |
"is_concept": true, |
|
|
503 |
"tuis_": "T016" |
|
|
504 |
}, |
|
|
505 |
{ |
|
|
506 |
"sent_i": 0, |
|
|
507 |
"norm_len": 4, |
|
|
508 |
"i_sent": 10, |
|
|
509 |
"lexspan": { |
|
|
510 |
"end": 62, |
|
|
511 |
"begin": 58 |
|
|
512 |
}, |
|
|
513 |
"is_wh": false, |
|
|
514 |
"i": 10, |
|
|
515 |
"is_superlative": false, |
|
|
516 |
"shape": 13110060611322374290, |
|
|
517 |
"is_stop": false, |
|
|
518 |
"dep": 7037928807040764755, |
|
|
519 |
"ent": 0, |
|
|
520 |
"norm": "lung", |
|
|
521 |
"pos_": "NOUN", |
|
|
522 |
"lemma_": "lung", |
|
|
523 |
"idx": 58, |
|
|
524 |
"tag": 15308085513773655218, |
|
|
525 |
"dep_": "compound", |
|
|
526 |
"is_punctuation": false, |
|
|
527 |
"ent_iob": 0, |
|
|
528 |
"tag_": "NN", |
|
|
529 |
"is_ent": false, |
|
|
530 |
"ent_iob_": "O", |
|
|
531 |
"children": [], |
|
|
532 |
"is_space": false, |
|
|
533 |
"ent_": "-<N>-", |
|
|
534 |
"shape_": "xxxx", |
|
|
535 |
"is_pronoun": false, |
|
|
536 |
"is_contraction": false, |
|
|
537 |
"cui": -1, |
|
|
538 |
"definition_": "-<N>-", |
|
|
539 |
"tuis": "frozenset()", |
|
|
540 |
"detected_name_": "-<N>-", |
|
|
541 |
"sub_names": [], |
|
|
542 |
"cui_": "-<N>-", |
|
|
543 |
"tui_descs_": "", |
|
|
544 |
"pref_name_": "-<N>-", |
|
|
545 |
"is_concept": false, |
|
|
546 |
"tuis_": "" |
|
|
547 |
}, |
|
|
548 |
{ |
|
|
549 |
"sent_i": 0, |
|
|
550 |
"norm_len": 6, |
|
|
551 |
"i_sent": 11, |
|
|
552 |
"lexspan": { |
|
|
553 |
"end": 69, |
|
|
554 |
"begin": 63 |
|
|
555 |
}, |
|
|
556 |
"is_wh": false, |
|
|
557 |
"i": 11, |
|
|
558 |
"is_superlative": false, |
|
|
559 |
"shape": 13110060611322374290, |
|
|
560 |
"is_stop": false, |
|
|
561 |
"dep": 416, |
|
|
562 |
"ent": 0, |
|
|
563 |
"norm": "cancer", |
|
|
564 |
"pos_": "NOUN", |
|
|
565 |
"lemma_": "cancer", |
|
|
566 |
"idx": 63, |
|
|
567 |
"tag": 15308085513773655218, |
|
|
568 |
"dep_": "dobj", |
|
|
569 |
"is_punctuation": false, |
|
|
570 |
"ent_iob": 0, |
|
|
571 |
"tag_": "NN", |
|
|
572 |
"is_ent": false, |
|
|
573 |
"ent_iob_": "O", |
|
|
574 |
"children": [ |
|
|
575 |
9, |
|
|
576 |
10 |
|
|
577 |
], |
|
|
578 |
"is_space": false, |
|
|
579 |
"ent_": "-<N>-", |
|
|
580 |
"shape_": "xxxx", |
|
|
581 |
"is_pronoun": false, |
|
|
582 |
"is_contraction": false, |
|
|
583 |
"cui": -1, |
|
|
584 |
"definition_": "-<N>-", |
|
|
585 |
"tuis": "frozenset()", |
|
|
586 |
"detected_name_": "-<N>-", |
|
|
587 |
"sub_names": [], |
|
|
588 |
"cui_": "-<N>-", |
|
|
589 |
"tui_descs_": "", |
|
|
590 |
"pref_name_": "-<N>-", |
|
|
591 |
"is_concept": false, |
|
|
592 |
"tuis_": "" |
|
|
593 |
}, |
|
|
594 |
{ |
|
|
595 |
"sent_i": 0, |
|
|
596 |
"norm_len": 5, |
|
|
597 |
"i_sent": 12, |
|
|
598 |
"lexspan": { |
|
|
599 |
"end": 75, |
|
|
600 |
"begin": 70 |
|
|
601 |
}, |
|
|
602 |
"is_wh": false, |
|
|
603 |
"i": 12, |
|
|
604 |
"is_superlative": false, |
|
|
605 |
"shape": 13110060611322374290, |
|
|
606 |
"is_stop": true, |
|
|
607 |
"dep": 423, |
|
|
608 |
"ent": 0, |
|
|
609 |
"norm": "while", |
|
|
610 |
"pos_": "SCONJ", |
|
|
611 |
"lemma_": "while", |
|
|
612 |
"idx": 70, |
|
|
613 |
"tag": 1292078113972184607, |
|
|
614 |
"dep_": "mark", |
|
|
615 |
"is_punctuation": false, |
|
|
616 |
"ent_iob": 0, |
|
|
617 |
"tag_": "IN", |
|
|
618 |
"is_ent": false, |
|
|
619 |
"ent_iob_": "O", |
|
|
620 |
"children": [], |
|
|
621 |
"is_space": false, |
|
|
622 |
"ent_": "-<N>-", |
|
|
623 |
"shape_": "xxxx", |
|
|
624 |
"is_pronoun": false, |
|
|
625 |
"is_contraction": false, |
|
|
626 |
"cui": -1, |
|
|
627 |
"definition_": "-<N>-", |
|
|
628 |
"tuis": "frozenset()", |
|
|
629 |
"detected_name_": "-<N>-", |
|
|
630 |
"sub_names": [], |
|
|
631 |
"cui_": "-<N>-", |
|
|
632 |
"tui_descs_": "", |
|
|
633 |
"pref_name_": "-<N>-", |
|
|
634 |
"is_concept": false, |
|
|
635 |
"tuis_": "" |
|
|
636 |
}, |
|
|
637 |
{ |
|
|
638 |
"sent_i": 0, |
|
|
639 |
"norm_len": 2, |
|
|
640 |
"i_sent": 13, |
|
|
641 |
"lexspan": { |
|
|
642 |
"end": 78, |
|
|
643 |
"begin": 76 |
|
|
644 |
}, |
|
|
645 |
"is_wh": false, |
|
|
646 |
"i": 13, |
|
|
647 |
"is_superlative": false, |
|
|
648 |
"shape": 4370460163704169311, |
|
|
649 |
"is_stop": true, |
|
|
650 |
"dep": 429, |
|
|
651 |
"ent": 0, |
|
|
652 |
"norm": "he", |
|
|
653 |
"pos_": "PRON", |
|
|
654 |
"lemma_": "he", |
|
|
655 |
"idx": 76, |
|
|
656 |
"tag": 13656873538139661788, |
|
|
657 |
"dep_": "nsubj", |
|
|
658 |
"is_punctuation": false, |
|
|
659 |
"ent_iob": 0, |
|
|
660 |
"tag_": "PRP", |
|
|
661 |
"is_ent": false, |
|
|
662 |
"ent_iob_": "O", |
|
|
663 |
"children": [], |
|
|
664 |
"is_space": false, |
|
|
665 |
"ent_": "-<N>-", |
|
|
666 |
"shape_": "xx", |
|
|
667 |
"is_pronoun": true, |
|
|
668 |
"is_contraction": false, |
|
|
669 |
"cui": -1, |
|
|
670 |
"definition_": "-<N>-", |
|
|
671 |
"tuis": "frozenset()", |
|
|
672 |
"detected_name_": "-<N>-", |
|
|
673 |
"sub_names": [], |
|
|
674 |
"cui_": "-<N>-", |
|
|
675 |
"tui_descs_": "", |
|
|
676 |
"pref_name_": "-<N>-", |
|
|
677 |
"is_concept": false, |
|
|
678 |
"tuis_": "" |
|
|
679 |
}, |
|
|
680 |
{ |
|
|
681 |
"sent_i": 0, |
|
|
682 |
"norm_len": 3, |
|
|
683 |
"i_sent": 14, |
|
|
684 |
"lexspan": { |
|
|
685 |
"end": 82, |
|
|
686 |
"begin": 79 |
|
|
687 |
}, |
|
|
688 |
"is_wh": false, |
|
|
689 |
"i": 14, |
|
|
690 |
"is_superlative": false, |
|
|
691 |
"shape": 4088098365541558500, |
|
|
692 |
"is_stop": true, |
|
|
693 |
"dep": 399, |
|
|
694 |
"ent": 0, |
|
|
695 |
"norm": "was", |
|
|
696 |
"pos_": "AUX", |
|
|
697 |
"lemma_": "be", |
|
|
698 |
"idx": 79, |
|
|
699 |
"tag": 17109001835818727656, |
|
|
700 |
"dep_": "advcl", |
|
|
701 |
"is_punctuation": false, |
|
|
702 |
"ent_iob": 0, |
|
|
703 |
"tag_": "VBD", |
|
|
704 |
"is_ent": false, |
|
|
705 |
"ent_iob_": "O", |
|
|
706 |
"children": [ |
|
|
707 |
12, |
|
|
708 |
13, |
|
|
709 |
15 |
|
|
710 |
], |
|
|
711 |
"is_space": false, |
|
|
712 |
"ent_": "-<N>-", |
|
|
713 |
"shape_": "xxx", |
|
|
714 |
"is_pronoun": false, |
|
|
715 |
"is_contraction": false, |
|
|
716 |
"cui": -1, |
|
|
717 |
"definition_": "-<N>-", |
|
|
718 |
"tuis": "frozenset()", |
|
|
719 |
"detected_name_": "-<N>-", |
|
|
720 |
"sub_names": [], |
|
|
721 |
"cui_": "-<N>-", |
|
|
722 |
"tui_descs_": "", |
|
|
723 |
"pref_name_": "-<N>-", |
|
|
724 |
"is_concept": false, |
|
|
725 |
"tuis_": "" |
|
|
726 |
}, |
|
|
727 |
{ |
|
|
728 |
"sent_i": 0, |
|
|
729 |
"norm_len": 2, |
|
|
730 |
"i_sent": 15, |
|
|
731 |
"lexspan": { |
|
|
732 |
"end": 85, |
|
|
733 |
"begin": 83 |
|
|
734 |
}, |
|
|
735 |
"is_wh": false, |
|
|
736 |
"i": 15, |
|
|
737 |
"is_superlative": false, |
|
|
738 |
"shape": 4370460163704169311, |
|
|
739 |
"is_stop": true, |
|
|
740 |
"dep": 443, |
|
|
741 |
"ent": 0, |
|
|
742 |
"norm": "in", |
|
|
743 |
"pos_": "ADP", |
|
|
744 |
"lemma_": "in", |
|
|
745 |
"idx": 83, |
|
|
746 |
"tag": 1292078113972184607, |
|
|
747 |
"dep_": "prep", |
|
|
748 |
"is_punctuation": false, |
|
|
749 |
"ent_iob": 0, |
|
|
750 |
"tag_": "IN", |
|
|
751 |
"is_ent": false, |
|
|
752 |
"ent_iob_": "O", |
|
|
753 |
"children": [ |
|
|
754 |
16 |
|
|
755 |
], |
|
|
756 |
"is_space": false, |
|
|
757 |
"ent_": "-<N>-", |
|
|
758 |
"shape_": "xx", |
|
|
759 |
"is_pronoun": false, |
|
|
760 |
"is_contraction": false, |
|
|
761 |
"cui": -1, |
|
|
762 |
"definition_": "-<N>-", |
|
|
763 |
"tuis": "frozenset()", |
|
|
764 |
"detected_name_": "-<N>-", |
|
|
765 |
"sub_names": [], |
|
|
766 |
"cui_": "-<N>-", |
|
|
767 |
"tui_descs_": "", |
|
|
768 |
"pref_name_": "-<N>-", |
|
|
769 |
"is_concept": false, |
|
|
770 |
"tuis_": "" |
|
|
771 |
}, |
|
|
772 |
{ |
|
|
773 |
"sent_i": 0, |
|
|
774 |
"norm_len": 7, |
|
|
775 |
"i_sent": 16, |
|
|
776 |
"lexspan": { |
|
|
777 |
"end": 93, |
|
|
778 |
"begin": 86 |
|
|
779 |
}, |
|
|
780 |
"is_wh": false, |
|
|
781 |
"i": 16, |
|
|
782 |
"is_superlative": false, |
|
|
783 |
"shape": 16072095006890171862, |
|
|
784 |
"is_stop": false, |
|
|
785 |
"dep": 439, |
|
|
786 |
"ent": 384, |
|
|
787 |
"norm": "Chicago", |
|
|
788 |
"pos_": "PROPN", |
|
|
789 |
"lemma_": "Chicago", |
|
|
790 |
"idx": 86, |
|
|
791 |
"tag": 15794550382381185553, |
|
|
792 |
"dep_": "pobj", |
|
|
793 |
"is_punctuation": false, |
|
|
794 |
"ent_iob": 3, |
|
|
795 |
"tag_": "NNP", |
|
|
796 |
"is_ent": true, |
|
|
797 |
"ent_iob_": "B", |
|
|
798 |
"children": [], |
|
|
799 |
"is_space": false, |
|
|
800 |
"ent_": "GPE", |
|
|
801 |
"shape_": "Xxxxx", |
|
|
802 |
"is_pronoun": false, |
|
|
803 |
"is_contraction": false, |
|
|
804 |
"cui": 8044, |
|
|
805 |
"definition_": "-<N>-", |
|
|
806 |
"tuis": [ |
|
|
807 |
"T083" |
|
|
808 |
], |
|
|
809 |
"detected_name_": "chicago", |
|
|
810 |
"sub_names": [ |
|
|
811 |
"chicago" |
|
|
812 |
], |
|
|
813 |
"cui_": "C0008044", |
|
|
814 |
"tui_descs_": "Geographic Area", |
|
|
815 |
"pref_name_": "chicago", |
|
|
816 |
"is_concept": true, |
|
|
817 |
"tuis_": "T083" |
|
|
818 |
}, |
|
|
819 |
{ |
|
|
820 |
"sent_i": 0, |
|
|
821 |
"norm_len": 1, |
|
|
822 |
"i_sent": 17, |
|
|
823 |
"lexspan": { |
|
|
824 |
"end": 94, |
|
|
825 |
"begin": 93 |
|
|
826 |
}, |
|
|
827 |
"is_wh": false, |
|
|
828 |
"i": 17, |
|
|
829 |
"is_superlative": false, |
|
|
830 |
"shape": 12646065887601541794, |
|
|
831 |
"is_stop": false, |
|
|
832 |
"dep": 445, |
|
|
833 |
"ent": 0, |
|
|
834 |
"norm": ".", |
|
|
835 |
"pos_": "PUNCT", |
|
|
836 |
"lemma_": ".", |
|
|
837 |
"idx": 93, |
|
|
838 |
"tag": 12646065887601541794, |
|
|
839 |
"dep_": "punct", |
|
|
840 |
"is_punctuation": true, |
|
|
841 |
"ent_iob": 0, |
|
|
842 |
"tag_": ".", |
|
|
843 |
"is_ent": false, |
|
|
844 |
"ent_iob_": "O", |
|
|
845 |
"children": [], |
|
|
846 |
"is_space": false, |
|
|
847 |
"ent_": "-<N>-", |
|
|
848 |
"shape_": ".", |
|
|
849 |
"is_pronoun": false, |
|
|
850 |
"is_contraction": false, |
|
|
851 |
"cui": -1, |
|
|
852 |
"definition_": "-<N>-", |
|
|
853 |
"tuis": "frozenset()", |
|
|
854 |
"detected_name_": "-<N>-", |
|
|
855 |
"sub_names": [], |
|
|
856 |
"cui_": "-<N>-", |
|
|
857 |
"tui_descs_": "", |
|
|
858 |
"pref_name_": "-<N>-", |
|
|
859 |
"is_concept": false, |
|
|
860 |
"tuis_": "" |
|
|
861 |
} |
|
|
862 |
] |
|
|
863 |
} |
|
|
864 |
] |
|
|
865 |
} |