File size: 47,076 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
{
    "paper_id": "2021",
    "header": {
        "generated_with": "S2ORC 1.0.0",
        "date_generated": "2023-01-19T13:27:29.873408Z"
    },
    "title": "Visualizing Cross-Lingual Discourse Relations in Multilingual TED Corpora",
    "authors": [
        {
            "first": "Zae",
            "middle": [
                "Myung"
            ],
            "last": "Kim",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "CNRS",
                "location": {
                    "region": "LIG"
                }
            },
            "email": ""
        },
        {
            "first": "Vassilina",
            "middle": [],
            "last": "Nikoulina",
            "suffix": "",
            "affiliation": {},
            "email": "vassilina.nikoulina@naverlabs.com"
        },
        {
            "first": "Dongyeop",
            "middle": [],
            "last": "Kang",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "University of Minnesota",
                "location": {}
            },
            "email": "dongyeop@umn.edu"
        },
        {
            "first": "Didier",
            "middle": [],
            "last": "Schwab",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "CNRS",
                "location": {
                    "region": "LIG"
                }
            },
            "email": "didier.schwab@univ-grenoble-alpes.fr"
        },
        {
            "first": "Laurent",
            "middle": [],
            "last": "Besacier",
            "suffix": "",
            "affiliation": {},
            "email": "laurent.besacier@naverlabs.com"
        },
        {
            "first": "Naver",
            "middle": [
                "Labs"
            ],
            "last": "Europe",
            "suffix": "",
            "affiliation": {},
            "email": ""
        }
    ],
    "year": "",
    "venue": null,
    "identifiers": {},
    "abstract": "This paper presents an interactive data dashboard that provides users with an overview of the preservation of discourse relations among 28 language pairs. We display a graph network depicting the cross-lingual discourse relations between a pair of languages for multilingual TED talks and provide a search function to look for sentences with specific keywords or relation types, facilitating ease of analysis on the cross-lingual discourse relations.",
    "pdf_parse": {
        "paper_id": "2021",
        "_pdf_hash": "",
        "abstract": [
            {
                "text": "This paper presents an interactive data dashboard that provides users with an overview of the preservation of discourse relations among 28 language pairs. We display a graph network depicting the cross-lingual discourse relations between a pair of languages for multilingual TED talks and provide a search function to look for sentences with specific keywords or relation types, facilitating ease of analysis on the cross-lingual discourse relations.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Abstract",
                "sec_num": null
            }
        ],
        "body_text": [
            {
                "text": "Discourse relations in texts describe how two discourse segments-which can be phrases, sentences, or even paragraphs-are connected logically to each other. For example, given the following two sentences, \"I have cooked some pasta.\" and \"I am hungry.\", we may infer that the latter is a cause for the former, in which case, the discourse relation between them can be classified as an implicit causation. 1 Now, when we translate these sentences into another language, the grammar and pragmatics of that language may prefer a certain discourse relation over the others that may not be the same as that of the original language. For example, in the case of Korean, which is an agglutinative language, the heavy usage of Korean postpositions (particles) dictates that the causal relation in the above example is revealed explicitly (regardless of the usage of explicit discourse connectives).",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Introduction",
                "sec_num": "1"
            },
            {
                "text": "Such a cross-lingual discourse analysis can provide some insights for improving the quality of (machine) translation (Meyer and Webber, 2013; Meyer and Pol\u00e1kov\u00e1, 2013; Guzm\u00e1n et al., 2014; Iruskieta et al., 2015; Chen et al., 2020) . This paper presents an interactive system that visualizes the cross-lingual discourse relations using two human-annotated multilingual discourse datasets (Zeyrek et al., 2019; Long et al., 2020) derived from TED talks following the Penn Discourse Treebank (PDTB) (Prasad et al., 2008) framework. Our interactive data dashboard provides users with an overview of relation preservation among all language pairs. We also display a graph network depicting the cross-lingual discourse relations between a pair of languages for a given TED talk, and provide a search function to find sentences with specific query or relation types. We believe that, with this visualization tool, users can easily browse through the multilingual talks and discover which parts of the talks share similar discourse relations, and where they diverge.",
                "cite_spans": [
                    {
                        "start": 117,
                        "end": 141,
                        "text": "(Meyer and Webber, 2013;",
                        "ref_id": "BIBREF6"
                    },
                    {
                        "start": 142,
                        "end": 167,
                        "text": "Meyer and Pol\u00e1kov\u00e1, 2013;",
                        "ref_id": "BIBREF5"
                    },
                    {
                        "start": 168,
                        "end": 188,
                        "text": "Guzm\u00e1n et al., 2014;",
                        "ref_id": "BIBREF2"
                    },
                    {
                        "start": 189,
                        "end": 212,
                        "text": "Iruskieta et al., 2015;",
                        "ref_id": "BIBREF3"
                    },
                    {
                        "start": 213,
                        "end": 231,
                        "text": "Chen et al., 2020)",
                        "ref_id": "BIBREF1"
                    },
                    {
                        "start": 388,
                        "end": 409,
                        "text": "(Zeyrek et al., 2019;",
                        "ref_id": "BIBREF11"
                    },
                    {
                        "start": 410,
                        "end": 428,
                        "text": "Long et al., 2020)",
                        "ref_id": "BIBREF4"
                    },
                    {
                        "start": 497,
                        "end": 518,
                        "text": "(Prasad et al., 2008)",
                        "ref_id": "BIBREF8"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Introduction",
                "sec_num": "1"
            },
            {
                "text": "We abbreviate the two discourse-annotated datasets, Zeyrek et al. (2019) and Long et al. (2020) , as TED-Mult and TED-ZH, respectively. TED-Mult involves seven languages (English, German, Lithuanian, Polish, Portuguese, Russian, and Turkish); and TED-ZH, two languages (English and Chinese). Thus, our analysis is conducted across the 28 possible language pairs.",
                "cite_spans": [
                    {
                        "start": 52,
                        "end": 72,
                        "text": "Zeyrek et al. (2019)",
                        "ref_id": "BIBREF11"
                    },
                    {
                        "start": 77,
                        "end": 95,
                        "text": "Long et al. (2020)",
                        "ref_id": "BIBREF4"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Preprocessing Datasets",
                "sec_num": "2"
            },
            {
                "text": "In order to facilitate cross-lingual analysis, we first performed sentence level alignment using a sentence segmentation tool 2 based on conditional random fields and a set of heuristic rules. We note that the reason for using this particular tool is that it guarantees the reconstruction of original texts when segmented sentences are joined together; and that it supports custom regex-based rules to add to allowlist or denylist if need be.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Preprocessing Datasets",
                "sec_num": "2"
            },
            {
                "text": "Once the TED talk scripts for all eight languages were segmented into sentence level, we cross-lingually matched the English sentences with sentences from other languages by the following procedure:",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Preprocessing Datasets",
                "sec_num": "2"
            },
            {
                "text": "1. Using a multilingual Sentence-BERT model Gurevych, 2019, 2020) , fixedsize sentence embeddings were computed for all sentences of the eight languages. 2. In addition, non-English sentences were machine-translated into English, and their English sentence embeddings were computed using a monolingual English Sentence-BERT model. 3. For each TED talk, cosine similarity scores were computed between English and non-English sentences using the multilingual sentences embeddings, and also the monolingual sentence embeddings, separately. 4. Similarly, chrF scores (Popovi\u0107, 2015) were computed between the English and the machine-translated (English) sentences. 5. We note that all of these scores were calculated for a fixed-size window of sentences as the cross-lingual sentences should be within a close range to each other. 6. Each type of scores was computed in both direction, i.e., English to Lang.X and Lang.X to English, and only their intersection of matched groups was kept. 7. Finally, the three types of scores were filtered by empirically defined thresholds and majority voting was used to pick the final matching groups of cross-lingual sentences for Lang.X:English pair for each TED document.",
                "cite_spans": [
                    {
                        "start": 44,
                        "end": 65,
                        "text": "Gurevych, 2019, 2020)",
                        "ref_id": null
                    },
                    {
                        "start": 563,
                        "end": 578,
                        "text": "(Popovi\u0107, 2015)",
                        "ref_id": "BIBREF7"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Preprocessing Datasets",
                "sec_num": "2"
            },
            {
                "text": "When aligning two non-English documents (Lang.X:Lang.Y), we used the alignments with the English (Lang.X:English and Lang.Y:English) as a pivot to cross-lingually match them.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Preprocessing Datasets",
                "sec_num": "2"
            },
            {
                "text": "To ensure the quality of the resulting alignments, we checked and revised each of them manually. We note that the manual correction was seldom required.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Preprocessing Datasets",
                "sec_num": "2"
            },
            {
                "text": "In total, we used seven cross-lingual talks from TED-Mult and four talks from TED-ZH, which are aligned with other languages. Overall, there are about 400 sentences per language, and Table 1 shows more details on the corpora statistics. ",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 183,
                        "end": 190,
                        "text": "Table 1",
                        "ref_id": "TABREF1"
                    }
                ],
                "eq_spans": [],
                "section": "Preprocessing Datasets",
                "sec_num": "2"
            },
            {
                "text": "Our system is implemented using Streamlit 3 which is an open-source web application framework in Python. The aligned dataset with codes 4 and the demo system 5 are available online.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "System Description",
                "sec_num": "3"
            },
            {
                "text": "The system is organized into three sections, \"Overall Patterns\", \"Pairwise Talks\", and \"Search\", where each section can be accessed via a navigation panel on the left-hand side.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "System Description",
                "sec_num": "3"
            },
            {
                "text": "3 https://streamlit.io/ 4 https://github.com/zaemyung/ Visualizing-Cross-Lingual-Discourse-Relations 5 https://share.streamlit.io/zaemyung/ visualizing-cross-lingual-discourserelations/main/main.py",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "System Description",
                "sec_num": "3"
            },
            {
                "text": "In PDTB-3 framework, a discourse relation is represented by four components: relation type, level-1, 2, and 3 senses. As shown in Table 2 , each subsequent sense level contains more fine-grained classes for the previous level, and we consider up to level-2 senses in our system. The overall pattern in the preservation of crosslingual discourse relations is depicted in a heatmap shown in Figure 1 . We compute how many relations are exactly matched across the language pairs, considering relation type and levels of senses. For each language pair, from left to right, each cell denotes the accuracy of matching (1) relation type (implicit or explicit); (2) level-1 senses; and (3) level-1 and 2 senses jointly. The accuracy of rela- We observe that the relation type and level-1 senses are quite well matched across the language pairs, averaging 77% and 68%, respectively. The accuracy scores drop when we compare more finegrained level-2 senses (48%). 6 We note that, while the scores for Figure 1 , were computed considering both the intra-and intersentential relations, it is possible to conduct the computation separately as well.",
                "cite_spans": [
                    {
                        "start": 954,
                        "end": 955,
                        "text": "6",
                        "ref_id": null
                    }
                ],
                "ref_spans": [
                    {
                        "start": 130,
                        "end": 137,
                        "text": "Table 2",
                        "ref_id": "TABREF2"
                    },
                    {
                        "start": 389,
                        "end": 397,
                        "text": "Figure 1",
                        "ref_id": "FIGREF1"
                    },
                    {
                        "start": 991,
                        "end": 999,
                        "text": "Figure 1",
                        "ref_id": "FIGREF1"
                    }
                ],
                "eq_spans": [],
                "section": "Overall Patterns",
                "sec_num": "3.1"
            },
            {
                "text": "Below the heatmap, we have the top-scoring rules mined from association rule mining (Agrawal et al., 1993) where the algorithm is applied to a collection of language-specific occurrences of relations for a given language pair. While most of the rules learned are \"identity\" (same discourse relation preserved across language pairs), there are some interesting non-identical rules as well. For example, it is often observed that with high confidence (0.59), level-1 sense Contingency for Russian co-occurs with level-1 sense Expansion for English. 7 6 We note that the avg. F1 score of matching Rhetorical Structure Theory (RST) relations across English, Spanish and Basque is 55% (Iruskieta et al., 2015) .",
                "cite_spans": [
                    {
                        "start": 84,
                        "end": 106,
                        "text": "(Agrawal et al., 1993)",
                        "ref_id": "BIBREF0"
                    },
                    {
                        "start": 549,
                        "end": 550,
                        "text": "6",
                        "ref_id": null
                    },
                    {
                        "start": 680,
                        "end": 704,
                        "text": "(Iruskieta et al., 2015)",
                        "ref_id": "BIBREF3"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Overall Patterns",
                "sec_num": "3.1"
            },
            {
                "text": "7 Both confidence and lift are two useful concepts in asso-At the end of the page, we present pie charts that illustrate the proportion of each relation type and level-1 and -2 senses of a source language being transferred to that of a target language, showing more fine-grained information than the heatmap. For example, in Figure 1 , the left pie chart for the level-1 senses (fist_cls) depicts that the English's Expansion sense is mostly preserved (78.4%) in the corresponding Russian sentences. Similarly, the right pie chart shows that the English's Implicit relation type is kept the same in the Russian sentences 80.9% of the time across all talks.",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 325,
                        "end": 333,
                        "text": "Figure 1",
                        "ref_id": "FIGREF1"
                    }
                ],
                "eq_spans": [],
                "section": "Overall Patterns",
                "sec_num": "3.1"
            },
            {
                "text": "These analyses on overall patterns can show how and where the discourse relations are preserved or diverged cross-lingually. From observations, we speculate about some of the causes for the divergence to be as follows: (1) The grammar and pragmatics of one language may favor the usage of one discourse relation over the other. (2) In some cases, especially when a discourse relation is Implicit, more than one relation could be possible. For ciation rule mining that select interesting rules from the set of all possible rules. Given a rule, A \u2192 B, the former (ranging from 0 to 1) computes the probability of seeing the consequent in a transaction given that it also contains the antecedent; the latter (ranging from 0 to \u221e), measures how much more often the antecedent and consequent of the rule occur jointly than if they were statistically independent (lif t = 1.0). example, we often observed cases where the Implicit.Expansion.Conjunction and Implicit.Comparison.Concession relations were quite similar to each other. 8 For each TED talk and a pair of source and target languages, an interactive graph network is rendered to show cross-lingual discourse relations ( Figure 2) . A node in the graph represents a sentence, while an edge depicts inter-sentential discourse relation between two nodes. In the graph of the screenshot, the orange graph represents the discourse graph for the first five sentences in the Russian document, while the blue one shows that of the English counterpart. The vertical line between the two discourse graphs indicates cross-lingual sentence alignment. We note that the context window of the generated graph can be adjusted by the Lower index and Upper index sliders, in the case of really long TED talks where, otherwise, the resulting graph would be too large to render efficiently.",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 1173,
                        "end": 1182,
                        "text": "Figure 2)",
                        "ref_id": "FIGREF2"
                    }
                ],
                "eq_spans": [],
                "section": "Overall Patterns",
                "sec_num": "3.1"
            },
            {
                "text": "With this graph representation, we can easily view which nodes share similar cross-lingual 8 These kinds of issues were resolved via majority voting or discussion when creating the annotated datasets by their respective authors. discourse relations.",
                "cite_spans": [
                    {
                        "start": 91,
                        "end": 92,
                        "text": "8",
                        "ref_id": null
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Pairwise Talks",
                "sec_num": "3.2"
            },
            {
                "text": "For example, in the screenshot, we observe that the second relation between the second and third sentences is Implicit.Expansion.Conjunction for Russian and Implicit.Expansion.Levelof-detail for English. In this case, for both languages, the relation types and level-1 senses are Implicit and Expansion, respectively, while the level-2 senses are different (Conjunction versus Level-of-detail). In addition, by clicking on the node, we can observe any intrasentential discourse relation present in the node by tagged spans in texts as illustrated in Figure 4 .",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 550,
                        "end": 558,
                        "text": "Figure 4",
                        "ref_id": "FIGREF4"
                    }
                ],
                "eq_spans": [],
                "section": "Pairwise Talks",
                "sec_num": "3.2"
            },
            {
                "text": "Users can look for a set of sentences that contain specific keywords or certain discourse relations on the \"Search\" page as displayed in Figure 3 . Currently, the search is based on an English keyword, and it retrieves all the other multilingual sentences that are cross-lingually aligned with the English ones containing the keyword. The results include both the previous and the subsequent neighboring sentences to show the context around the query sentence.",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 137,
                        "end": 145,
                        "text": "Figure 3",
                        "ref_id": "FIGREF3"
                    }
                ],
                "eq_spans": [],
                "section": "Search",
                "sec_num": "3.3"
            },
            {
                "text": "We presented an interactive system that visualizes the cross-lingual discourse relations among multilingual TED talks, along with other related statistics. As future work, we plan to integrate an online multilingual discourse parser and a multilingual machine translation model into the system so that users can translate and analyze any given document and its translation on the fly.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Conclusion",
                "sec_num": "4"
            },
            {
                "text": "It is implicit because these sentences are not connected with an explicit discourse marker such as \"Because\".",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "https://github.com/zaemyung/sentsplit",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            }
        ],
        "back_matter": [],
        "bib_entries": {
            "BIBREF0": {
                "ref_id": "b0",
                "title": "Mining association rules between sets of items in large databases",
                "authors": [
                    {
                        "first": "Rakesh",
                        "middle": [],
                        "last": "Agrawal",
                        "suffix": ""
                    },
                    {
                        "first": "Tomasz",
                        "middle": [],
                        "last": "Imieli\u0144ski",
                        "suffix": ""
                    },
                    {
                        "first": "Arun",
                        "middle": [],
                        "last": "Swami",
                        "suffix": ""
                    }
                ],
                "year": 1993,
                "venue": "Proceedings of the 1993 ACM SIGMOD international conference on Management of data",
                "volume": "",
                "issue": "",
                "pages": "207--216",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Rakesh Agrawal, Tomasz Imieli\u0144ski, and Arun Swami. 1993. Mining association rules between sets of items in large databases. In Proceedings of the 1993 ACM SIGMOD international conference on Manage- ment of data, pages 207-216.",
                "links": null
            },
            "BIBREF1": {
                "ref_id": "b1",
                "title": "Modeling discourse structure for document-level neural machine translation",
                "authors": [
                    {
                        "first": "Junxuan",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    },
                    {
                        "first": "Xiang",
                        "middle": [],
                        "last": "Li",
                        "suffix": ""
                    },
                    {
                        "first": "Jiarui",
                        "middle": [],
                        "last": "Zhang",
                        "suffix": ""
                    },
                    {
                        "first": "Chulun",
                        "middle": [],
                        "last": "Zhou",
                        "suffix": ""
                    },
                    {
                        "first": "Jianwei",
                        "middle": [],
                        "last": "Cui",
                        "suffix": ""
                    },
                    {
                        "first": "Bin",
                        "middle": [],
                        "last": "Wang",
                        "suffix": ""
                    },
                    {
                        "first": "Jinsong",
                        "middle": [],
                        "last": "Su",
                        "suffix": ""
                    }
                ],
                "year": 2020,
                "venue": "Proceedings of the First Workshop on Automatic Simultaneous Translation",
                "volume": "",
                "issue": "",
                "pages": "30--36",
                "other_ids": {
                    "DOI": [
                        "10.18653/v1/2020.autosimtrans-1.5"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Junxuan Chen, Xiang Li, Jiarui Zhang, Chulun Zhou, Jianwei Cui, Bin Wang, and Jinsong Su. 2020. Mod- eling discourse structure for document-level neural machine translation. In Proceedings of the First Workshop on Automatic Simultaneous Translation, pages 30-36, Seattle, Washington. Association for Computational Linguistics.",
                "links": null
            },
            "BIBREF2": {
                "ref_id": "b2",
                "title": "Using discourse structure improves machine translation evaluation",
                "authors": [
                    {
                        "first": "Francisco",
                        "middle": [],
                        "last": "Guzm\u00e1n",
                        "suffix": ""
                    },
                    {
                        "first": "Shafiq",
                        "middle": [],
                        "last": "Joty",
                        "suffix": ""
                    },
                    {
                        "first": "Llu\u00eds",
                        "middle": [],
                        "last": "M\u00e0rquez",
                        "suffix": ""
                    },
                    {
                        "first": "Preslav",
                        "middle": [],
                        "last": "Nakov",
                        "suffix": ""
                    }
                ],
                "year": 2014,
                "venue": "Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics",
                "volume": "1",
                "issue": "",
                "pages": "687--698",
                "other_ids": {
                    "DOI": [
                        "10.3115/v1/P14-1065"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Francisco Guzm\u00e1n, Shafiq Joty, Llu\u00eds M\u00e0rquez, and Preslav Nakov. 2014. Using discourse structure im- proves machine translation evaluation. In Proceed- ings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Pa- pers), pages 687-698, Baltimore, Maryland. Associ- ation for Computational Linguistics.",
                "links": null
            },
            "BIBREF3": {
                "ref_id": "b3",
                "title": "A qualitative comparison method for rhetorical structures: identifying different discourse structures in multilingual corpora. Language resources and evaluation",
                "authors": [
                    {
                        "first": "Mikel",
                        "middle": [],
                        "last": "Iruskieta",
                        "suffix": ""
                    },
                    {
                        "first": "Maite",
                        "middle": [],
                        "last": "Iria Da Cunha",
                        "suffix": ""
                    },
                    {
                        "first": "",
                        "middle": [],
                        "last": "Taboada",
                        "suffix": ""
                    }
                ],
                "year": 2015,
                "venue": "",
                "volume": "49",
                "issue": "",
                "pages": "263--309",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Mikel Iruskieta, Iria Da Cunha, and Maite Taboada. 2015. A qualitative comparison method for rhetor- ical structures: identifying different discourse struc- tures in multilingual corpora. Language resources and evaluation, 49(2):263-309.",
                "links": null
            },
            "BIBREF4": {
                "ref_id": "b4",
                "title": "Shallow discourse annotation for Chinese TED talks",
                "authors": [
                    {
                        "first": "Wanqiu",
                        "middle": [],
                        "last": "Long",
                        "suffix": ""
                    },
                    {
                        "first": "Xinyi",
                        "middle": [],
                        "last": "Cai",
                        "suffix": ""
                    },
                    {
                        "first": "James",
                        "middle": [],
                        "last": "Reid",
                        "suffix": ""
                    },
                    {
                        "first": "Bonnie",
                        "middle": [],
                        "last": "Webber",
                        "suffix": ""
                    },
                    {
                        "first": "Deyi",
                        "middle": [],
                        "last": "Xiong",
                        "suffix": ""
                    }
                ],
                "year": 2020,
                "venue": "Proceedings of the 12th Language Resources and Evaluation Conference",
                "volume": "",
                "issue": "",
                "pages": "1025--1032",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Wanqiu Long, Xinyi Cai, James Reid, Bonnie Web- ber, and Deyi Xiong. 2020. Shallow discourse an- notation for Chinese TED talks. In Proceedings of the 12th Language Resources and Evaluation Con- ference, pages 1025-1032, Marseille, France. Euro- pean Language Resources Association.",
                "links": null
            },
            "BIBREF5": {
                "ref_id": "b5",
                "title": "Machine translation with many manually labeled discourse connectives",
                "authors": [
                    {
                        "first": "Thomas",
                        "middle": [],
                        "last": "Meyer",
                        "suffix": ""
                    },
                    {
                        "first": "Lucie",
                        "middle": [],
                        "last": "Pol\u00e1kov\u00e1",
                        "suffix": ""
                    }
                ],
                "year": 2013,
                "venue": "Proceedings of the Workshop on Discourse in Machine Translation",
                "volume": "",
                "issue": "",
                "pages": "43--50",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Thomas Meyer and Lucie Pol\u00e1kov\u00e1. 2013. Machine translation with many manually labeled discourse connectives. In Proceedings of the Workshop on Dis- course in Machine Translation, pages 43-50, Sofia, Bulgaria. Association for Computational Linguis- tics.",
                "links": null
            },
            "BIBREF6": {
                "ref_id": "b6",
                "title": "Implicitation of discourse connectives in (machine) translation",
                "authors": [
                    {
                        "first": "Thomas",
                        "middle": [],
                        "last": "Meyer",
                        "suffix": ""
                    },
                    {
                        "first": "Bonnie",
                        "middle": [],
                        "last": "Webber",
                        "suffix": ""
                    }
                ],
                "year": 2013,
                "venue": "Proceedings of the Workshop on Discourse in Machine Translation",
                "volume": "",
                "issue": "",
                "pages": "19--26",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Thomas Meyer and Bonnie Webber. 2013. Implicita- tion of discourse connectives in (machine) transla- tion. In Proceedings of the Workshop on Discourse in Machine Translation, pages 19-26, Sofia, Bul- garia. Association for Computational Linguistics.",
                "links": null
            },
            "BIBREF7": {
                "ref_id": "b7",
                "title": "chrF: character n-gram F-score for automatic MT evaluation",
                "authors": [
                    {
                        "first": "Maja",
                        "middle": [],
                        "last": "Popovi\u0107",
                        "suffix": ""
                    }
                ],
                "year": 2015,
                "venue": "Proceedings of the Tenth Workshop on Statistical Machine Translation",
                "volume": "",
                "issue": "",
                "pages": "392--395",
                "other_ids": {
                    "DOI": [
                        "10.18653/v1/W15-3049"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Maja Popovi\u0107. 2015. chrF: character n-gram F-score for automatic MT evaluation. In Proceedings of the Tenth Workshop on Statistical Machine Translation, pages 392-395, Lisbon, Portugal. Association for Computational Linguistics.",
                "links": null
            },
            "BIBREF8": {
                "ref_id": "b8",
                "title": "The Penn Discourse TreeBank 2.0",
                "authors": [
                    {
                        "first": "Rashmi",
                        "middle": [],
                        "last": "Prasad",
                        "suffix": ""
                    },
                    {
                        "first": "Nikhil",
                        "middle": [],
                        "last": "Dinesh",
                        "suffix": ""
                    },
                    {
                        "first": "Alan",
                        "middle": [],
                        "last": "Lee",
                        "suffix": ""
                    },
                    {
                        "first": "Eleni",
                        "middle": [],
                        "last": "Miltsakaki",
                        "suffix": ""
                    },
                    {
                        "first": "Livio",
                        "middle": [],
                        "last": "Robaldo",
                        "suffix": ""
                    },
                    {
                        "first": "Aravind",
                        "middle": [],
                        "last": "Joshi",
                        "suffix": ""
                    },
                    {
                        "first": "Bonnie",
                        "middle": [],
                        "last": "Webber",
                        "suffix": ""
                    }
                ],
                "year": 2008,
                "venue": "Proceedings of the Sixth International Conference on Language Resources and Evaluation (LREC'08)",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Rashmi Prasad, Nikhil Dinesh, Alan Lee, Eleni Milt- sakaki, Livio Robaldo, Aravind Joshi, and Bon- nie Webber. 2008. The Penn Discourse TreeBank 2.0. In Proceedings of the Sixth International Conference on Language Resources and Evaluation (LREC'08), Marrakech, Morocco. European Lan- guage Resources Association (ELRA).",
                "links": null
            },
            "BIBREF9": {
                "ref_id": "b9",
                "title": "Sentencebert: Sentence embeddings using siamese bertnetworks",
                "authors": [
                    {
                        "first": "Nils",
                        "middle": [],
                        "last": "Reimers",
                        "suffix": ""
                    },
                    {
                        "first": "Iryna",
                        "middle": [],
                        "last": "Gurevych",
                        "suffix": ""
                    }
                ],
                "year": 2019,
                "venue": "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Nils Reimers and Iryna Gurevych. 2019. Sentence- bert: Sentence embeddings using siamese bert- networks. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing. Association for Computational Linguistics.",
                "links": null
            },
            "BIBREF10": {
                "ref_id": "b10",
                "title": "Making monolingual sentence embeddings multilingual using knowledge distillation",
                "authors": [
                    {
                        "first": "Nils",
                        "middle": [],
                        "last": "Reimers",
                        "suffix": ""
                    },
                    {
                        "first": "Iryna",
                        "middle": [],
                        "last": "Gurevych",
                        "suffix": ""
                    }
                ],
                "year": 2020,
                "venue": "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing. Association for Computational Linguistics",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Nils Reimers and Iryna Gurevych. 2020. Making monolingual sentence embeddings multilingual us- ing knowledge distillation. In Proceedings of the 2020 Conference on Empirical Methods in Natu- ral Language Processing. Association for Computa- tional Linguistics.",
                "links": null
            },
            "BIBREF11": {
                "ref_id": "b11",
                "title": "Ted multilingual discourse bank (tedmdb): A parallel corpus annotated in the pdtb style",
                "authors": [
                    {
                        "first": "Deniz",
                        "middle": [],
                        "last": "Zeyrek",
                        "suffix": ""
                    },
                    {
                        "first": "Am\u00e1lia",
                        "middle": [],
                        "last": "Mendes",
                        "suffix": ""
                    },
                    {
                        "first": "Yulia",
                        "middle": [],
                        "last": "Grishina",
                        "suffix": ""
                    },
                    {
                        "first": "Murathan",
                        "middle": [],
                        "last": "Kurfal\u0131",
                        "suffix": ""
                    },
                    {
                        "first": "Samuel",
                        "middle": [],
                        "last": "Gibbon",
                        "suffix": ""
                    },
                    {
                        "first": "Maciej",
                        "middle": [],
                        "last": "Ogrodniczuk",
                        "suffix": ""
                    }
                ],
                "year": 2019,
                "venue": "Language Resources and Evaluation",
                "volume": "",
                "issue": "",
                "pages": "1--27",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Deniz Zeyrek, Am\u00e1lia Mendes, Yulia Grishina, Mu- rathan Kurfal\u0131, Samuel Gibbon, and Maciej Ogrod- niczuk. 2019. Ted multilingual discourse bank (ted- mdb): A parallel corpus annotated in the pdtb style. Language Resources and Evaluation, pages 1-27.",
                "links": null
            }
        },
        "ref_entries": {
            "FIGREF1": {
                "text": "A screenshot showing the \"Overall Patterns\" of discourse relations among language pairs.",
                "uris": null,
                "num": null,
                "type_str": "figure"
            },
            "FIGREF2": {
                "text": "A screenshot showing the cross-lingual relation network graph of \"Pairwise Talks\". tions for cross-lingual alignments is computed as: 2 * |I\u2229J| |I|+|J| , where I and J are the set of corresponding relations for each alignment pair.",
                "uris": null,
                "num": null,
                "type_str": "figure"
            },
            "FIGREF3": {
                "text": "A screenshot showing the results of \"Search\" function where a sentence containing a specific English query is returned along its immediate neighboring sentences.",
                "uris": null,
                "num": null,
                "type_str": "figure"
            },
            "FIGREF4": {
                "text": "A screenshot showing an intra-sentential relation within a node.",
                "uris": null,
                "num": null,
                "type_str": "figure"
            },
            "TABREF1": {
                "content": "<table/>",
                "text": "Table of corpora statistics for TED-Mult and TED-ZH.",
                "num": null,
                "type_str": "table",
                "html": null
            },
            "TABREF2": {
                "content": "<table/>",
                "text": "",
                "num": null,
                "type_str": "table",
                "html": null
            }
        }
    }
}