-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathiwpt2017.html
581 lines (525 loc) · 22.1 KB
/
iwpt2017.html
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
<!DOCTYPE html>
<html xmlns="http://www.w3.org/1999/xhtml">
<head>
<meta charset="utf-8" />
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<meta name="generator" content="pandoc" />
<meta name="author" content="Niko Partanen" />
<title>Discussion on manually annotated and parsed Komi dependencies</title>
<script src="lib/jquery-1.11.3/jquery.min.js"></script>
<meta name="viewport" content="width=device-width, initial-scale=1" />
<link href="lib/bootstrap-3.3.5/css/bootstrap.min.css" rel="stylesheet" />
<script src="lib/bootstrap-3.3.5/js/bootstrap.min.js"></script>
<script src="lib/bootstrap-3.3.5/shim/html5shiv.min.js"></script>
<script src="lib/bootstrap-3.3.5/shim/respond.min.js"></script>
<script src="lib/navigation-1.1/tabsets.js"></script>
<link href="lib/highlightjs-1.1/default.css" rel="stylesheet" />
<script src="lib/highlightjs-1.1/highlight.js"></script>
<meta http-equiv="content-type" content="text/html; charset=utf-8"/>
<title>Tree visualization</title>
<link rel="stylesheet" type="text/css" href="css/jquery-ui-redmond.css"/>
<link rel="stylesheet" type="text/css" href="css/style.css"/>
<link rel="stylesheet" type="text/css" href="css/style-vis.css"/>
<script type="text/javascript" src="head.load.min.js"></script>
<!-- <link rel="shortcut icon" href="favicon.ico"/> -->
<!-- Leipzig.js -->
<!-- CSS
<link rel="stylesheet" href="//cdn.jsdelivr.net/leipzig/latest/leipzig.min.css">-->
<!-- JavaScript
<script src="//cdn.jsdelivr.net/leipzig/latest/leipzig.min.js"></script> -->
<style type="text/css">code{white-space: pre;}</style>
<style type="text/css">
pre:not([class]) {
background-color: white;
}
</style>
<script type="text/javascript">
if (window.hljs && document.readyState && document.readyState === "complete") {
window.setTimeout(function() {
hljs.initHighlighting();
}, 0);
}
</script>
<style type="text/css">
h1 {
font-size: 34px;
}
h1.title {
font-size: 38px;
}
h2 {
font-size: 30px;
}
h3 {
font-size: 24px;
}
h4 {
font-size: 18px;
}
h5 {
font-size: 16px;
}
h6 {
font-size: 12px;
}
.table th:not([align]) {
text-align: left;
}
</style>
</head>
<body>
<style type = "text/css">
.main-container {
max-width: 940px;
margin-left: auto;
margin-right: auto;
}
code {
color: inherit;
background-color: rgba(0, 0, 0, 0.04);
}
img {
max-width:100%;
height: auto;
}
.tabbed-pane {
padding-top: 12px;
}
button.code-folding-btn:focus {
outline: none;
}
</style>
<div class="container-fluid main-container">
<!-- tabsets -->
<script>
$(document).ready(function () {
window.buildTabsets("TOC");
});
</script>
<!-- code folding -->
<div class="fluid-row" id="header">
<h1 class="title toc-ignore">Discussion on manually annotated and parsed Komi dependencies</h1>
<h4 class="author"><em>Niko Partanen</em></h4>
<h4 class="date"><em>7/5/2017</em></h4>
</div>
<h2 id="section"></h2>
<ul>
<li>Manually annotated utterance <strong>is on top</strong></li>
<li>One below comes from our BIST-parser</li>
<li><strong>Both</strong> human and machine generated versions may contain mistakes ;)</li>
</ul>
<p>Suggestions, improvements and comments can be sent to me in email <a href="mailto:[email protected]">[email protected]</a>.</p>
<p>So what is going on here is that we have been training multilingual dependency parser for Komi and Northern Saami, with the idea that we can use multilingual word embeddings and UD corpora from different languages as our training data. By controlling the training data it is possible to adjust the model. For example, the result drifts quite a lot depending from ratios of Finnish and Russian in a model being 20/80 or 50/50 and so on. Of course Komi is in no way a mixture of Finnish and Russian, but it shares similarities with both of them, in one hand for being a related language, and with Russian there is the long language contact that makes them more similar.</p>
<p>One problem is that there is a possibility that the test data contains mistakes in itself, as I’m not any kind of expert in Universal Dependency style annotations, and had never done anything with them before. Actually reporting errors would be most appreciated.</p>
<p>Good news is that because we have the output files all saved on our server, we can very easily recalculate the scores if there are changes in the evaluation data. I already noticed several mistakes, and I think there are several cases where I have misunderstood the UD annotation model.</p>
<h2 id="analysis-of-sentences">Analysis of sentences</h2>
<p>So as explained above, the first one is the ground truth sentence and the second one is generated based on Russian treebanks:</p>
<p><div class="conllu-parse" tabs="yes">1 А а SCONJ _ _ 2 cc
2 сэсся сэсся ADV _ _ 5 advmod
3 пыр пыр ADV _ _ 4 advmod
4 отаджыкӧсь ота ADJ _ _ 5 amod
5 лоисны лоны VERB _ _ 0 root
6 . . PUNCT _ _ 5 punct
</div> <div class="conllu-parse" tabs="yes">1 А а SCONJ _ _ 5 mark
2 сэсся сэсся ADV _ _ 4 advmod
3 пыр пыр ADV _ _ 4 advmod
4 отаджыкӧсь ота ADJ _ _ 5 advcl
5 лоисны лоны VERB _ _ 0 root
6 . . PUNCT _ _ 5 punct
</div></p>
<p></br></p>
<p>I don’t know why it thinks <code>А</code> is of category <a href="http://universaldependencies.org/u/dep/mark.html">mark</a>. Now when I read the description that doesn’t even look that wrong actually…</p>
<p>With other example it is obvious that the subject is already encoded wrong in the test data.</p>
<p><div class="conllu-parse" tabs="yes">1 Кыліс кывны VERB _ _ 0 root
2 зэв зэв ADV _ _ 3 advmod
3 ылысянь ылысянь ADV _ _ 1 obl
4 муркӧдчӧм муркӧдчыны VERB _ _ 5 acl
5 шы шы NOUN _ _ 1 nsubj
6 . . PUNCT _ _ 5 punct
</div> <div class="conllu-parse" tabs="yes">1 Кыліс кывны VERB _ _ 0 root
2 зэв зэв ADV _ _ 4 advmod
3 ылысянь ылысянь ADV _ _ 4 advmod
4 муркӧдчӧм муркӧдчыны VERB _ _ 1 advcl
5 шы шы NOUN _ _ 1 obl
6 . . PUNCT _ _ 5 punct
</div></p>
<p></br></p>
<p>Here is another one which is analysed correctly. The sentence is very simple <em>[the water] [calmed down]</em>.</p>
<p><div class="conllu-parse" tabs="yes">1 Ваыс ва NOUN _ _ 2 nsubj
2 лӧнис лӧньны VERB _ _ 0 root
3 . . PUNCT _ _ 2 punct
</div> <div class="conllu-parse" tabs="yes">1 Ваыс ва NOUN _ _ 2 nsubj
2 лӧнис лӧньны VERB _ _ 0 root
3 . . PUNCT _ _ 2 punct
</div></p>
<p></br></p>
<p>Here is another one where the analysis is same, the sentence being bit more complex <em>[the fish][entirely][got quiet]</em>:</p>
<p><div class="conllu-parse" tabs="yes">1 Чериыд чери NOUN _ _ 3 nsubj
2 дзикӧдз дзикӧдз ADV _ _ 3 advmod
3 раммис раммыны VERB _ _ 0 root
4 . . PUNCT _ _ 3 punct
</div> <div class="conllu-parse" tabs="yes">1 Чериыд чери NOUN _ _ 3 nsubj
2 дзикӧдз дзикӧдз ADV _ _ 3 advmod
3 раммис раммыны VERB _ _ 0 root
4 . . PUNCT _ _ 3 punct
</div></p>
<p></br></p>
<p>Here also result is good. The sentence is <strong>[even][the tail][(it) didn’t][show]</strong></p>
<p><div class="conllu-parse" tabs="yes">1 Весиг весиг ADV _ _ 2 advmod
2 бӧжсӧ бӧж NOUN _ _ 4 obj
3 эз оз AUX _ _ 4 aux
4 петкӧдлы петкӧдлыны VERB _ _ 0 root
5 . . PUNCT _ _ 4 punct
</div> <div class="conllu-parse" tabs="yes">1 Весиг весиг ADV _ _ 2 advmod
2 бӧжсӧ бӧж NOUN _ _ 4 obl
3 эз оз AUX _ _ 4 aux
4 петкӧдлы петкӧдлыны VERB _ _ 0 root
5 . . PUNCT _ _ 4 punct
</div></p>
<p></br></p>
<p>In this case it is absolutely clear that I didn’t know very well how to annotate the postpositions, so maybe the Russian model understood those better after all? <strong>Note: compare this kind of situations to some Finnish examples</strong>.</p>
<p><div class="conllu-parse" tabs="yes">1 Кымӧрыс кымӧр NOUN _ _ 5 nsubj
2 недыр недыр ADV _ _ 3 advmod
3 повзьӧдчӧм повзьӧдчыны VERB _ _ 4 advmod
4 бӧрын бӧр ADP _ _ 1 obl
5 вешйис вешйыны VERB _ _ 0 root
6 бокӧ бокӧ ADP _ _ 5 obl
7 . . PUNCT _ _ 6 punct
</div> <div class="conllu-parse" tabs="yes">1 Кымӧрыс кымӧр NOUN _ _ 3 nsubj
2 недыр недыр ADV _ _ 3 advmod
3 повзьӧдчӧм повзьӧдчыны VERB _ _ 0 root
4 бӧрын бӧр ADP _ _ 5 case
5 вешйис вешйыны VERB _ _ 3 obl
6 бокӧ бокӧ ADP _ _ 5 case
7 . . PUNCT _ _ 5 punct
</div></p>
<p></br></p>
<p>Also here the question is more if two consecutive adverbs should be linked to one another or directly to the root, as it is in the generated case.</p>
<p><div class="conllu-parse" tabs="yes">1 Коркӧ коркӧ ADV _ _ 4 advmod
2 век век ADV _ _ 3 advmod
3 нин нин ADV _ _ 4 advmod
4 гымалас гымавны VERB _ _ 0 root
5 и и CCONJ _ _ 6 cc
6 зэрас зэрны VERB _ _ 4 conj
7 . . PUNCT _ _ 6 punct
</div> <div class="conllu-parse" tabs="yes">1 Коркӧ коркӧ ADV _ _ 4 advmod
2 век век ADV _ _ 4 advmod
3 нин нин ADV _ _ 4 advmod
4 гымалас гымавны VERB _ _ 0 root
5 и и CCONJ _ _ 6 cc
6 зэрас зэрны VERB _ _ 4 conj
7 . . PUNCT _ _ 6 punct
</div></p>
<p></br></p>
<p>Here the only mistake is in object not being detected correctly.</p>
<p><div class="conllu-parse" tabs="yes">1 Ньӧжйӧник ньӧжйӧник ADV _ _ 2 advmod
2 чукӧрті чукӧртны VERB _ _ 0 root
3 чери·кыян·кӧлуйӧс чери·кыян·кӧлуй NOUN _ _ 2 obj
4 да да CCONJ _ _ 5 cc
5 мӧдӧдчи мӧдӧдчыны VERB _ _ 2 conj
6 гортлань горт NOUN _ _ 5 obl
7 . . PUNCT _ _ 6 punct
</div> <div class="conllu-parse" tabs="yes">1 Ньӧжйӧник ньӧжйӧник ADV _ _ 2 advmod
2 чукӧрті чукӧртны VERB _ _ 0 root
3 чери·кыян·кӧлуйӧс чери·кыян·кӧлуй NOUN _ _ 2 obl
4 да да CCONJ _ _ 5 cc
5 мӧдӧдчи мӧдӧдчыны VERB _ _ 2 conj
6 гортлань горт NOUN _ _ 5 obl
7 . . PUNCT _ _ 6 punct
</div></p>
<p></br></p>
<p>This repeats the same question on adverbs.</p>
<p><div class="conllu-parse" tabs="yes">1 Сійӧ сійӧ PRON _ _ 4 nsubj
2 вочасӧн вочасӧн ADV _ _ 3 advmod
3 пыр пыр ADV _ _ 4 advmod
4 паськаліс паськавны VERB _ _ 0 root
5 да да CCONJ _ _ 6 cc
6 лӧзӧдіс лӧзӧдны VERB _ _ 4 conj
7 . . PUNCT _ _ 6 punct
</div> <div class="conllu-parse" tabs="yes">1 Сійӧ сійӧ PRON _ _ 4 nsubj
2 вочасӧн вочасӧн ADV _ _ 4 advmod
3 пыр пыр ADV _ _ 4 advmod
4 паськаліс паськавны VERB _ _ 0 root
5 да да CCONJ _ _ 6 cc
6 лӧзӧдіс лӧзӧдны VERB _ _ 4 xcomp
7 . . PUNCT _ _ 6 punct
</div></p>
<p></br></p>
<p>This kind of mistake with objects and obliques we have seen many times already.</p>
<p><div class="conllu-parse" tabs="yes">1 Ӧддзӧді ӧддзӧдны VERB _ _ 0 root
2 воськовъясӧс воськов NOUN _ _ 1 obj
3 . . PUNCT _ _ 2 punct
</div> <div class="conllu-parse" tabs="yes">1 Ӧддзӧді ӧддзӧдны VERB _ _ 0 root
2 воськовъясӧс воськов NOUN _ _ 1 obl
3 . . PUNCT _ _ 2 punct
</div></p>
<p></br></p>
<p>Here is another correctly analysed one:</p>
<p><div class="conllu-parse" tabs="yes">1 Но но CCONJ _ _ 3 cc
2 кымӧрыд кымӧр NOUN _ _ 3 nsubj
3 локтіс локны VERB _ _ 0 root
4 нӧшта нӧшта ADV _ _ 5 advmod
5 ӧдйӧджык ӧдйӧ ADV _ _ 3 advmod
6 . . PUNCT _ _ 5 punct
</div> <div class="conllu-parse" tabs="yes">1 Но но CCONJ _ _ 3 cc
2 кымӧрыд кымӧр NOUN _ _ 3 nsubj
3 локтіс локны VERB _ _ 0 root
4 нӧшта нӧшта ADV _ _ 5 advmod
5 ӧдйӧджык ӧдйӧ ADV _ _ 3 advmod
6 . . PUNCT _ _ 5 punct
</div></p>
<p></br></p>
<p>This is also correct.</p>
<p><div class="conllu-parse" tabs="yes">1 Кутіс кутны VERB _ _ 0 root
2 нин нин ADV _ _ 3 advmod
3 гымавны-чардавны гымавны-чардавны VERB _ _ 1 xcomp
4 . . PUNCT _ _ 3 punct
</div> <div class="conllu-parse" tabs="yes">1 Кутіс кутны VERB _ _ 0 root
2 нин нин ADV _ _ 3 advmod
3 гымавны-чардавны гымавны-чардавны VERB _ _ 1 xcomp
4 . . PUNCT _ _ 3 punct
</div></p>
<p></br></p>
<p>This was is not correct, but I think the word order is the reason here. Subject is in the very end. <code>А</code> is again annotated as <code>mark</code>, maybe this is the correct annotation then?</p>
<p><div class="conllu-parse" tabs="yes">1 А а SCONJ _ _ 4 cc
2 сэсся сэсся ADV _ _ 4 advmod
3 вылісянь вылісянь ADV _ _ 4 obl
4 мӧдӧдчисны мӧдӧдчыны VERB _ _ 0 root
5 гырысь гырысь ADJ _ _ 6 amod
6 зэр·войтъяс зэр·войт NOUN _ _ 4 nsubj
7 . . PUNCT _ _ 6 punct
</div> <div class="conllu-parse" tabs="yes">1 А а SCONJ _ _ 4 mark
2 сэсся сэсся ADV _ _ 4 advmod
3 вылісянь вылісянь ADV _ _ 4 advmod
4 мӧдӧдчисны мӧдӧдчыны VERB _ _ 0 root
5 гырысь гырысь ADJ _ _ 6 amod
6 зэр·войтъяс зэр·войт NOUN _ _ 4 obl
7 . . PUNCT _ _ 4 punct
</div></p>
<p></br></p>
<p>Here is similar situation as before with advmods after one another. The analysis tries to suggest subject in the beginning, but here the subject is not marked. Are the subjects more compulsory in Russian than Komi? Similarly I wasn’t that sure whether we should annotate the verb there as copula or what. Is <code>aux</code> or <code>xcomp</code> better?</p>
<p><div class="conllu-parse" tabs="yes">1 Зэрсьыс зэр NOUN _ _ 2 nmod
2 дзебсьыны дзебсьыны VERB _ _ 0 root
3 вӧлі вӧвны VERB _ _ 2 cop
4 весьшӧрӧ весьшӧрӧ ADV _ _ 3 advmod
5 нин нин ADV _ _ 4 advmod
6 . . PUNCT _ _ 5 punct
</div> <div class="conllu-parse" tabs="yes">1 Зэрсьыс зэр NOUN _ _ 2 nsubj
2 дзебсьыны дзебсьыны VERB _ _ 0 root
3 вӧлі вӧвны VERB _ _ 2 xcomp
4 весьшӧрӧ весьшӧрӧ ADV _ _ 2 advmod
5 нин нин ADV _ _ 2 advmod
6 . . PUNCT _ _ 5 punct
</div> </br></p>
<p>Here is a gerund <strong>while it was raining</strong>. Wasn’t that sure how to annotate it.</p>
<p><div class="conllu-parse" tabs="yes">1 Зэригас зэрны VERB _ _ 2 advcl
2 кая кайны VERB _ _ 0 root
3 гортӧдз горт NOUN _ _ 2 obl
4 . . PUNCT _ _ 3 punct
</div> <div class="conllu-parse" tabs="yes">1 Зэригас зэрны VERB _ _ 0 root
2 кая кайны VERB _ _ 1 xcomp
3 гортӧдз горт NOUN _ _ 2 obj
4 . . PUNCT _ _ 3 punct
</div></p>
<p></br></p>
<p>Here the negation verb is somehow analysed as punctuation? Where this can come from?</p>
<p><div class="conllu-parse" tabs="yes">1 Эг оз AUX _ _ 4 aux
2 нин нин ADV _ _ 4 advmod
3 сэтшӧма сэтшӧма ADV _ _ 4 advmod
4 кут кутны VERB _ _ 0 root
5 тэрмасьны тэрмасьны VERB _ _ 4 xcomp
6 . . PUNCT _ _ 5 punct
</div> <div class="conllu-parse" tabs="yes">1 Эг оз AUX _ _ 4 punct
2 нин нин ADV _ _ 4 advmod
3 сэтшӧма сэтшӧма ADV _ _ 4 advmod
4 кут кутны VERB _ _ 0 root
5 тэрмасьны тэрмасьны VERB _ _ 4 xcomp
6 . . PUNCT _ _ 5 punct
</div></p>
<p></br></p>
<p>Here is a typo in the test data, рудӧдны should be <code>xcomp</code> of кутіс. But it doesn’t effect the fact that subject is analysed from in the machine generated output. Is the clause around the subject somehow different from Russian constructions?</p>
<p><div class="conllu-parse" tabs="yes">1 Тӧв·ныркӧд тӧв·ныр NOUN _ _ 2 nmod
2 тшӧтш тшӧтш ADV _ _ 3 advmod
3 сынӧдыс сынӧд NOUN _ _ 4 nsubj
4 кутіс кутны VERB _ _ 0 root
5 рудӧдны рудӧдны VERB _ _ 3 xcomp
6 . . PUNCT _ _ 5 punct
</div> <div class="conllu-parse" tabs="yes">1 Тӧв·ныркӧд тӧв·ныр NOUN _ _ 4 nsubj
2 тшӧтш тшӧтш ADV _ _ 3 advmod
3 сынӧдыс сынӧд NOUN _ _ 1 nmod
4 кутіс кутны VERB _ _ 0 root
5 рудӧдны рудӧдны VERB _ _ 4 xcomp
6 . . PUNCT _ _ 5 punct
</div></p>
<p></br></p>
<p>Here are my mistakes with the adverbs, but also the subject is not found correctly.</p>
<p><div class="conllu-parse" tabs="yes">1 Тайӧ тайӧ PRON _ _ 2 det
2 рудыс руд ADJ _ _ 5 nsubj
3 зэв зэв ADV _ _ 4 advmod
4 ӧдйӧ ӧдйӧ ADV _ _ 5 advmod
5 матыстчис матыстчыны VERB _ _ 0 root
6 мелань ме PRON _ _ 5 obl
7 . . PUNCT _ _ 6 punct
</div> <div class="conllu-parse" tabs="yes">1 Тайӧ тайӧ PRON _ _ 5 nsubj
2 рудыс руд ADJ _ _ 5 obl
3 зэв зэв ADV _ _ 5 advmod
4 ӧдйӧ ӧдйӧ ADV _ _ 5 advmod
5 матыстчис матыстчыны VERB _ _ 0 root
6 мелань ме PRON _ _ 5 obl
7 . . PUNCT _ _ 6 punct
</div></p>
<p></br></p>
<p>This one is also a case where the subject is in the very end. Or can be analyzed so?</p>
<p><div class="conllu-parse" tabs="yes">1 Медбӧрын медбӧрын ADV _ _ 3 advmod
2 чужӧмӧ чужӧм NOUN _ _ 3 obl
3 кучкисны кучкыны VERB _ _ 0 root
4 йи йи NOUN _ _ 5 nmod
5 торъяс тор NOUN _ _ 3 nsubj
6 . . PUNCT _ _ 5 punct
</div> <div class="conllu-parse" tabs="yes">1 Медбӧрын медбӧрын ADV _ _ 3 advmod
2 чужӧмӧ чужӧм NOUN _ _ 3 obl
3 кучкисны кучкыны VERB _ _ 0 root
4 йи йи NOUN _ _ 3 obl
5 торъяс тор NOUN _ _ 4 nmod
6 . . PUNCT _ _ 5 punct
</div></p>
<p></br></p>
<p>Have to think about this one…</p>
<p><div class="conllu-parse" tabs="yes">1 Восьлавны восьлавны VERB _ _ 5 xcomp
2 водзӧ водз ADP _ _ 1 advmod
3 эг оз AUX _ _ 5 aux
4 нин нин ADV _ _ 5 advmod
5 кут кутны VERB _ _ 0 root
6 вермыны вермыны VERB _ _ 5 xcomp
7 . . PUNCT _ _ 6 punct
</div> <div class="conllu-parse" tabs="yes">1 Восьлавны восьлавны VERB _ _ 0 root
2 водзӧ водз ADP _ _ 3 case
3 эг оз AUX _ _ 5 aux
4 нин нин ADV _ _ 5 advmod
5 кут кутны VERB _ _ 1 parataxis
6 вермыны вермыны VERB _ _ 5 xcomp
7 . . PUNCT _ _ 6 punct
</div></p>
<p></br></p>
<p>Here also analysis is different. Maybe one thing is that in Russian would never have just one noun alone in a position like this?</p>
<p><div class="conllu-parse" tabs="yes">1 Тӧлыс тӧв NOUN _ _ 4 nsubj
2 да да CCONJ _ _ 3 cc
3 шерыс шер NOUN _ _ 1 conj
4 пӧрӧдісны пӧрӧдны VERB _ _ 0 root
5 кок·йывсьыд кок·йыв NOUN _ _ 4 obl
6 . . PUNCT _ _ 5 punct
</div> <div class="conllu-parse" tabs="yes">1 Тӧлыс тӧв NOUN _ _ 4 nsubj
2 да да CCONJ _ _ 3 cc
3 шерыс шер NOUN _ _ 1 nmod
4 пӧрӧдісны пӧрӧдны VERB _ _ 0 root
5 кок·йывсьыд кок·йыв NOUN _ _ 4 advmod
6 . . PUNCT _ _ 5 punct
</div></p>
<script type="text/javascript">
// TODO pull this stuff from a more reasonable location
var root = 'https://nikopartanen.github.io/conllu-vis/';
head.js(
// External libraries
root + 'lib/ext/jquery.min.js',
root + 'lib/ext/jquery.svg.min.js',
root + 'lib/ext/jquery.svgdom.min.js',
root + 'lib/ext/jquery-ui.min.js',
root + 'lib/ext/waypoints.min.js',
// brat helper modules
root + 'lib/brat/configuration.js',
root + 'lib/brat/util.js',
root + 'lib/brat/annotation_log.js',
root + 'lib/ext/webfont.js',
// brat modules
root + 'lib/brat/dispatcher.js',
root + 'lib/brat/url_monitor.js',
root + 'lib/brat/visualizer.js',
// embedding support modules
'https://nikopartanen.github.io/conllu-vis/lib/local/annodoc.js',
root + 'conllu.js'
//root + 'lib/local/config.js'
);
var bratCollData = {
'entity_types': [
// this is optional
{
'type': 'SPAN_DEFAULT',
'bgColor': '#7fa2ff',
'borderColor': 'darken'
},
{
'type': 'ARC_DEFAULT',
'color': 'black',
'arrowHead': 'triangle,5,5',
'labelArrow': 'triangle,3,8',
'args': [
{
'role': 'arg1',
'targets': [ 'token' ]
},
{
'role': 'arg2',
'targets': [ 'token' ]
}
]
},
{
'type': 'token',
'labels': [ '\u00A0\u00A0' ], // non-breaking space for empty
},
{
'type': '-',
'labels': [ '\u00A0\u00A0' ], // non-breaking space for empty
}
],
'event_attribute_types': [],
'entity_attribute_types': [
{
'type': 'Name',
'values': {
'Name' : { 'glyph': '(N)' },
},
},
],
'relation_types': [
// this is optional
],
'event_types': [],
};
var webFontURLs = [
// root + 'static/fonts/Astloch-Bold.ttf',
root + 'static/fonts/PT_Sans-Caption-Web-Regular.ttf',
root + 'static/fonts/Liberation_Sans-Regular.ttf'
];
/* collection data, not applicable here. */
var documentCollections = {};
head.ready(function() {
// performes all embedding and support functions
Annodoc.activate(bratCollData, documentCollections);
});
</script>
<div id="footer">
<p class="footer-text">embedded brat © 2010- the <a href="http://brat.nlplab.org/about.html" style="color:gray">brat contributors</a></p>
</div>
</div>
</div>
<script>
// add bootstrap table styles to pandoc tables
function bootstrapStylePandocTables() {
$('tr.header').parent('thead').parent('table').addClass('table table-condensed');
}
$(document).ready(function () {
bootstrapStylePandocTables();
});
</script>
<!-- dynamically load mathjax for compatibility with self-contained -->
<script>
(function () {
var script = document.createElement("script");
script.type = "text/javascript";
script.src = "https://mathjax.rstudio.com/latest/MathJax.js?config=TeX-AMS-MML_HTMLorMML";
document.getElementsByTagName("head")[0].appendChild(script);
})();
</script>
</body>
</html>