File size: 51,203 Bytes
6fa4bc9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
{
    "paper_id": "S13-1012",
    "header": {
        "generated_with": "S2ORC 1.0.0",
        "date_generated": "2023-01-19T15:42:21.892106Z"
    },
    "title": "PolyUCOMP-CORE TYPED: Computing Semantic Textual Similarity using Overlapped Senses",
    "authors": [
        {
            "first": "Jian",
            "middle": [],
            "last": "Xu",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "The Hong Kong Polytechnic University Department of Computing Hung Hom",
                "location": {
                    "settlement": "Kowloon, Hong Kong"
                }
            },
            "email": ""
        },
        {
            "first": "Qin",
            "middle": [],
            "last": "Lu",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "The Hong Kong Polytechnic University Department of Computing Hung Hom",
                "location": {
                    "settlement": "Kowloon, Hong Kong"
                }
            },
            "email": ""
        }
    ],
    "year": "",
    "venue": null,
    "identifiers": {},
    "abstract": "The Semantic Textual Similarity (STS) task aims to exam the degree of semantic equivalence between sentences (Agirre et al., 2012). This paper presents the work of the Hong Kong Polytechnic University (PolyUCOMP) team which has participated in the STS core and typed tasks of SemEval-2013. For the STS core task, the PolyUCOMP system disambiguates words senses using contexts and then determine sentence similarity by counting the number of senses they shared. For the STS typed task, the string kernel (Lodhi et al., 2002) is used to compute similarity between two entities to avoid string variations in entities.",
    "pdf_parse": {
        "paper_id": "S13-1012",
        "_pdf_hash": "",
        "abstract": [
            {
                "text": "The Semantic Textual Similarity (STS) task aims to exam the degree of semantic equivalence between sentences (Agirre et al., 2012). This paper presents the work of the Hong Kong Polytechnic University (PolyUCOMP) team which has participated in the STS core and typed tasks of SemEval-2013. For the STS core task, the PolyUCOMP system disambiguates words senses using contexts and then determine sentence similarity by counting the number of senses they shared. For the STS typed task, the string kernel (Lodhi et al., 2002) is used to compute similarity between two entities to avoid string variations in entities.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Abstract",
                "sec_num": null
            }
        ],
        "body_text": [
            {
                "text": "Sentence similarity computation plays an important role in text summarization and social network applications (Erkan et al., 2004; Jin et al., 2011) . The SemEval 2012 competition initiated a task targeted at Semantic Textual Similarity (STS) between sentence pairs (Agirre et al., 2012) . Given a set of sentence pairs, participants are required to assign to each sentence pair a similarity score.",
                "cite_spans": [
                    {
                        "start": 110,
                        "end": 130,
                        "text": "(Erkan et al., 2004;",
                        "ref_id": "BIBREF3"
                    },
                    {
                        "start": 131,
                        "end": 148,
                        "text": "Jin et al., 2011)",
                        "ref_id": "BIBREF6"
                    },
                    {
                        "start": 266,
                        "end": 287,
                        "text": "(Agirre et al., 2012)",
                        "ref_id": "BIBREF1"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Introduction",
                "sec_num": "1"
            },
            {
                "text": "Because a sentence has only a limited amount of content words, it is difficult to determine sentence similarities. To solve this problem, Hatzivassiloglou et al. (1999) proposed to use linguistic features as indicators of text similarity to address the problem of sparse representation of sentences. Mihalcea et al. (2006) measured sentence similarity using component words in sentences. Li et al. (2006) proposed to incorporate the semantic vector and word order to calculate sentence similarity. Biemann et al.",
                "cite_spans": [
                    {
                        "start": 138,
                        "end": 168,
                        "text": "Hatzivassiloglou et al. (1999)",
                        "ref_id": "BIBREF9"
                    },
                    {
                        "start": 300,
                        "end": 322,
                        "text": "Mihalcea et al. (2006)",
                        "ref_id": "BIBREF7"
                    },
                    {
                        "start": 388,
                        "end": 404,
                        "text": "Li et al. (2006)",
                        "ref_id": "BIBREF10"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Introduction",
                "sec_num": "1"
            },
            {
                "text": "(2012) applied the log-linear regression model by combining the simple string based measures, for example, word ngrams and semantic similarity measures, for example, textual entailment. Similarly, Saric et al. (2012) used a support vector regression model which incorporates features computed from sentence pairs. The features are knowledge-and corpus-based word similarity, ngram overlaps, WordNet augmented word overlap, syntactic features and so on. Xu et al. (2012) combined semantic vectors with skip bigrams to determine sentence similarity, whereas the skip bigrams take into the sequential order between words.",
                "cite_spans": [
                    {
                        "start": 197,
                        "end": 216,
                        "text": "Saric et al. (2012)",
                        "ref_id": "BIBREF2"
                    },
                    {
                        "start": 453,
                        "end": 469,
                        "text": "Xu et al. (2012)",
                        "ref_id": "BIBREF5"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Introduction",
                "sec_num": "1"
            },
            {
                "text": "In our approach to the STS task, words in sentences are assigned with appropriate senses using their contexts. Sentence similarity is computed by calculating the number of shared senses in both sentences since it is reasonable to assume that similar sentences should have more overlapping senses.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Introduction",
                "sec_num": "1"
            },
            {
                "text": "For the STS-TYPED task, variations might occur in author names, people involved, time expression and location. Thus, string kernel is applied to compute similarity between entities because it can capture variations between entities. Moreover, for the event similarity in STS-TYPED task, semantic relatedness between verbs is derived the WordNet.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Introduction",
                "sec_num": "1"
            },
            {
                "text": "The rest of this paper is structured as follows. Section 2 describes sentence similarity using sense overlapping and string kernel. Section 3 gives the performance evaluation. Section 4 is the conclusion.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Introduction",
                "sec_num": "1"
            },
            {
                "text": "Words are used to convey meaning in a sentence. They are tagged with appropriate senses initially and then sentence similarity is calculated based on the number of shared senses.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Similarity between Sentences",
                "sec_num": "2"
            },
            {
                "text": "When comparing word features, we did not compare their surface equality, but we first conceptualize these words and then calculate their similarities based on the hierarchial structure in WordNet. For a word in a sentence, it will be assigned a WordNet sense. In this paper, we focus on the Word Sense Disambiguation (WSD) algorithm taken by Banerjee and Pederson (2003) . They measured the semantic relatedness between concepts by counting the shared words in their WordNet glosses.",
                "cite_spans": [
                    {
                        "start": 342,
                        "end": 370,
                        "text": "Banerjee and Pederson (2003)",
                        "ref_id": null
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Sense Overlapping",
                "sec_num": "2.1"
            },
            {
                "text": "In WordNet, a word sense is represented by a synset which has a gloss that defines the concept that it represents. For example, the words walking, afoot, ambulate constitute a single synset which has gloss representations as follows, walking: the act of traveling by foot afoot: traveling by foot ambulate: walk about To lift the limitations of dictionary glosses which are fairly short with insufficient vocabulary, we utilize the glosses of related senses since we assume that words co-occur in one sentence share related senses and the more glosses two senses share, the more similar they are. Therefore, we extract not only glosses of target synset, but also the glosses of the hypernym, hyponym, meronym, holonym and troponym synsets of the target synset to form a synset context. Finally, we compare the sentence contexts with different synset contexts to determine which sense should be assigned to the words.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Sense Overlapping",
                "sec_num": "2.1"
            },
            {
                "text": "To disambiguate word senses, a window of contexts surrounding the the target word is specified and a set of candidate word senses are extracted for the content word (noun, verb, adjective) within that window. Let the current target word index i = 0 that is, w 0 , the window size be 2n+1 and \u2212n \u2264 i \u2264 +n. Let |w i | be the number of senses for word w i and the j th sense of w i is s i,j , where 1 \u2264 j \u2264 |w i |. Next is to assign an appropriate sense k to the target word. We achieve this by adding together the relatedness scores calculated by comparing the senses of the target word and senses of every non-target word within the window of context. The sense score for the current target word w 0 is defined as,",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Sense Overlapping",
                "sec_num": "2.1"
            },
            {
                "text": "Sense k = n i=\u2212n |w i | j=1 relatedness(s 0,k , s i,j ) (1)",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Sense Overlapping",
                "sec_num": "2.1"
            },
            {
                "text": "The k th sense which has the biggest sense score will be chosen as the right sense for the target word w 0 . Now remains the question of how to define the relatedness between two synsets. It is defined as,",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Sense Overlapping",
                "sec_num": "2.1"
            },
            {
                "text": "relatedness(s 0,k , s i,j ) = score(gloss(s 0,k ), gloss(s i,j )) +score(hype(s 0,k ), hype(s i,j )) +score(hypo(s 0,k ), hypo(s i,j )) +score(hype(s 0,k ), gloss(s i,j )) +score(gloss(s 0,k ), hype(s i,j )) (2)",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Sense Overlapping",
                "sec_num": "2.1"
            },
            {
                "text": "In Equation 2, the score function counts the number of overlapping words between two glosses. However, if there is a phrasal n-word overlap, then a score of n 2 will be assigned, thus encouraging the longer n-word overlap. Let V denote the set of nword overlaps shared between two glosses, the score is defined as,",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Sense Overlapping",
                "sec_num": "2.1"
            },
            {
                "text": "EQUATION",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [
                    {
                        "start": 0,
                        "end": 8,
                        "text": "EQUATION",
                        "ref_id": "EQREF",
                        "raw_str": "score = w\u2208V w 2",
                        "eq_num": "(3)"
                    }
                ],
                "section": "Sense Overlapping",
                "sec_num": "2.1"
            },
            {
                "text": "where w refers to the number of words in w. In so doing, we can have corresponding senses for the sentence Castro celebrates 86th birthday Monday as follows, castro/10886929-n celebrate/02490877-v birthday/15250178-n monday/15163979-n To find the n-word overlap, we found that contiguous words in two glosses lie in the diagonal of a matrix, take the senses walk and afoot for example, their glosses are, walking: the act of traveling by foot afoot: traveling by foot Place the walking glosses in rows and afoot glosses in columns, we get the matrix representation in Figure 1 , Figure 1 shows that travel by foot is a continuous sequence of words shared by two glosses. Steps to find n-word overlapping are:",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 568,
                        "end": 576,
                        "text": "Figure 1",
                        "ref_id": "FIGREF0"
                    },
                    {
                        "start": 579,
                        "end": 587,
                        "text": "Figure 1",
                        "ref_id": "FIGREF0"
                    }
                ],
                "eq_spans": [],
                "section": "Sense Overlapping",
                "sec_num": "2.1"
            },
            {
                "text": "(1) Construct a matrix for two sentences;",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Sense Overlapping",
                "sec_num": "2.1"
            },
            {
                "text": "(2) Get continuous n-word overlapping, n is greater than 1;",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Sense Overlapping",
                "sec_num": "2.1"
            },
            {
                "text": "(3) Set the cell values to 0 if they are contained in continuous n-word.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Sense Overlapping",
                "sec_num": "2.1"
            },
            {
                "text": "(4) Get the words (unigrams) which are shared by two sentences.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Sense Overlapping",
                "sec_num": "2.1"
            },
            {
                "text": "Take By the step 2, we will get the b c and its corresponding cells cell(1,0) and cell(2,1). We then set the two cells to zero, and obtain an updated matrix as follows, b c a d a 0 0 1 0 b 0 0 0 0 c 0 0 0 0 d 0 0 0 1 In Table 2 , we found that cell(0,2) and cell(3,3) have values greater than zero. Therefore, a and b will be extracted the common terms.",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 220,
                        "end": 227,
                        "text": "Table 2",
                        "ref_id": "TABREF1"
                    }
                ],
                "eq_spans": [],
                "section": "Sense Overlapping",
                "sec_num": "2.1"
            },
            {
                "text": "This approach can also be applied to find common n-word overlaps between sentences, for example, s 1 : Olli Heinonen, the Head of the International Atomic Energy Agency delegation to Iran, declared yesterday that the agency has reached an agreement with Tehran on the method of conducting the negotiations pertaining to its nuclear program.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Sense Overlapping",
                "sec_num": "2.1"
            },
            {
                "text": "s 2 : leader of international atomic energy agency delegation to iran , olli heinonen said yesterday , that the agency concluded a mutual understanding with tehran on the way to manage talks depending upon its atomic program .",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Sense Overlapping",
                "sec_num": "2.1"
            },
            {
                "text": "We will have ngrams with n ranging from 1 to 7, such as, unigram: of, to, its, program, yesterday bigram: olli heinonen trigram: that the agency four-gram: with tehran on the seven-gram: international atomic energy agency delegation to iran Similarity between two sentences is calculated by counting the number of overlapped n-words. The similarity for s 1 and s 2 is, (1 + 1 + 1 + 1 + 1) + (2) 2 + (3) 2 + (4) 2 + (7) 2 = 83.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Sense Overlapping",
                "sec_num": "2.1"
            },
            {
                "text": "For the STS-TYPED task, when comparing whether people or authors are similar or not, we found that some entity mentions may have tiny variations, for example,",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "String kernel",
                "sec_num": "2.2"
            },
            {
                "text": "The difference between the entities lies in fact that the second entity has one more dot. In this case, string kernel would be a good choice in verifying they are similar or not. If we consider n=2, we obtain 79-dimensional feature space where the two entities are mapped in Table 3 .",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 275,
                        "end": 282,
                        "text": "Table 3",
                        "ref_id": "TABREF2"
                    }
                ],
                "eq_spans": [],
                "section": "E Vincent Harris and E.Vincent Harris",
                "sec_num": null
            },
            {
                "text": "In Table 3 , \u03bb is the decay factor, in the range of [0, 1] , that penalizes the longer distance of a subsequence. Formally, string kernel is defined as, where n is the set of all possible subsequences of length n. u indicates an item in the set, for example, the subsequence ev in Table 3 . \u03c6 u (s) is the feature mapping of the subsequences in s. In so doing, we can have similarity between entities in Table 3 as follows:",
                "cite_spans": [
                    {
                        "start": 52,
                        "end": 55,
                        "text": "[0,",
                        "ref_id": null
                    },
                    {
                        "start": 56,
                        "end": 58,
                        "text": "1]",
                        "ref_id": null
                    }
                ],
                "ref_spans": [
                    {
                        "start": 3,
                        "end": 10,
                        "text": "Table 3",
                        "ref_id": "TABREF2"
                    },
                    {
                        "start": 281,
                        "end": 288,
                        "text": "Table 3",
                        "ref_id": "TABREF2"
                    },
                    {
                        "start": 404,
                        "end": 411,
                        "text": "Table 3",
                        "ref_id": "TABREF2"
                    }
                ],
                "eq_spans": [],
                "section": "E Vincent Harris and E.Vincent Harris",
                "sec_num": null
            },
            {
                "text": "K n (s, t) = u\u2208 n \u03c6 u (s) \u2022 \u03c6 u (t) (4) ev ei en \u2022 \u2022 \u2022 e. \u2022 \u2022 \u2022 rs is \u03c6(evincentharris) \u03bb 2 \u03bb 3 + \u03bb 13 \u03bb 2 + \u03bb 4 + \u03bb 7 \u2022 \u2022 \u2022 0 \u2022 \u2022 \u2022 \u03bb 3 + \u03bb 4 \u03bb 2 + \u03bb 12 \u03c6(e.vincentharris) \u03bb 3 \u03bb 4 + \u03bb 14 \u03bb 2 + \u03bb 5 + \u03bb 8 \u2022 \u2022 \u2022 \u03bb 2 \u2022 \u2022 \u2022 \u03bb 3 + \u03bb 4 \u03bb 2 + \u03bb 12",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "E Vincent Harris and E.Vincent Harris",
                "sec_num": null
            },
            {
                "text": "K n (s, t) = \u03bb 2 \u00d7 \u03bb 3 + (\u03bb 3 + \u03bb 13 ) \u00d7 (\u03bb 4 + \u03bb 14 ) + \u2022 \u2022 \u2022+(\u03bb 3 +\u03bb 4 )\u00d7(\u03bb 3 +\u03bb 4 )+(\u03bb 2 +\u03bb 12 )\u00d7(\u03bb 2 +\u03bb 12 )",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "E Vincent Harris and E.Vincent Harris",
                "sec_num": null
            },
            {
                "text": "To avoid enumeration of all subsequences for similarity measurement, dynamic programming, similar to the method by Lodhi et al. (2002) is used here for similarity calculation.",
                "cite_spans": [
                    {
                        "start": 115,
                        "end": 134,
                        "text": "Lodhi et al. (2002)",
                        "ref_id": "BIBREF4"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "E Vincent Harris and E.Vincent Harris",
                "sec_num": null
            },
            {
                "text": "The STS-CORE task is to quantify how similar two sentences are. We simply use the sense overlapping approach to compute the similarity. Since this approach needs to find appropriate senses for each word based on its contexts. The number of contextual words is set to 5. Experiments are conducted on four datasets.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Experiments",
                "sec_num": "3"
            },
            {
                "text": "They are: headlines mined from news sources by European Media, OnWN extracted from from WordNet and OntoNotes, FNWN from WordNet and FrameNet and SMT dataset from DARPA GALE HTER and HyTER. The results of our system (PolyUCOMP-RUN1) are given in Table 4 , Our system achieves rather lower performance in the OnWN and FNWN datasets. This is because it is difficult to use contextual terms to find the correct senses for words in sentences of these two datasets. Take the two sentences in OnWN dataset for example, s 1 : the act of choosing among alternatives s 2 : the act of changing one thing for another thing.",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 246,
                        "end": 255,
                        "text": "Table 4 ,",
                        "ref_id": "TABREF3"
                    }
                ],
                "eq_spans": [],
                "section": "Experiments",
                "sec_num": "3"
            },
            {
                "text": "The valid concepts for the two sentences are:",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Experiments",
                "sec_num": "3"
            },
            {
                "text": "c 1 : 06532095-n 05790944-n c 2 :",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Experiments",
                "sec_num": "3"
            },
            {
                "text": "00030358-n 00126264-v 00002452-n 00002452-n c 1 and c 2 have no shared senses, resulting in a zero similarity between s 1 and s 2 . However, s 1 and s 2 should have the same meaning. Moreover, in the FNWN dataset, the sentence lengths are unbalanced, for example, s 1 : there exist a number of different possible events that may happen in the future. in most cases, there is an agent involved who has to consider which of the possible events will or should occur. a salient entity which is deeply involved in the event may also be mentioned.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Experiments",
                "sec_num": "3"
            },
            {
                "text": "s 2 : doing as one pleases or chooses;",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Experiments",
                "sec_num": "3"
            },
            {
                "text": "s 1 has 48 tokens with punctuations being excluded and s 2 has only 6 tokens. This would affect our system performance as well.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Experiments",
                "sec_num": "3"
            },
            {
                "text": "For the STS-TYPED task, data set is taken from Europeana, which provides millions of books, paintings, films, museum objects and archival records that have been digitised throughout Europe. Each item has one line per type, where the type can be the title of a record, list of subject terms, textual description of the record, creator of the record and date of the record. Participating systems are supposed to compute similarities between semistructured items. In this task, we take the strategies in Table 5 , Jaccard denotes the Jaccard similarity measure. Stringkernel + Jaccard means that two types are similar if they share many terms, for example, location; and string kernel is used to determine whether two locations are similar or not. For the type of event, we extract verbs from records and count the number of shared verbs between two records. The verb similarity is obtained through WordNet. The general similarity is equal to the average of the 7 scores. Also, Stanford CoreNLP tool 1 is used to extract author, date, time, location and handle part-of-speech tagging. In this STS-TYPED task, we use string kernel and WordNet to determine whether two terms are similar and increase the number of counts if their similarity exceeds a certain threshold. Therefore, we have chosen 0.4, 0.5 and 0.6 in a heuristic manner and obtained three different runs. Experimental results are given in Table 6 .",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 501,
                        "end": 508,
                        "text": "Table 5",
                        "ref_id": "TABREF6"
                    },
                    {
                        "start": 1399,
                        "end": 1406,
                        "text": "Table 6",
                        "ref_id": "TABREF5"
                    }
                ],
                "eq_spans": [],
                "section": "Experiments",
                "sec_num": "3"
            },
            {
                "text": "Since the types of author, subject and description are not related to either string kernel or WordNet, their performances remain unchanged during three runs.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Experiments",
                "sec_num": "3"
            },
            {
                "text": "In the Semantic Textual Similarity task of SemEval-2013, to capture the meaning between sentences, we proposed to disambiguate word senses using contexts and then determine sentence similarity by counting the senses they shared. First, word senses are disambiguated by means of the contextual 1 http://nlp.stanford.edu/software/corenlp.shtml words. When determining similarity between two senses (synsets), n-word overlapping approach is used for counting the number of shared words in two glosses. Besides, string kernel is used to capture similarity between entities to avoid variations between entities. Our approach is simple and we will apply regression models to determine sentence similarity on the basis of these features in future work.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Conclusions and Future Work",
                "sec_num": "4"
            }
        ],
        "back_matter": [],
        "bib_entries": {
            "BIBREF0": {
                "ref_id": "b0",
                "title": "UKP: Computing Semantic Textual Similarity by Combining Multiple Content Similarity Measures",
                "authors": [
                    {
                        "first": "B",
                        "middle": [],
                        "last": "Daniel",
                        "suffix": ""
                    },
                    {
                        "first": "Chris",
                        "middle": [],
                        "last": "Biemann",
                        "suffix": ""
                    },
                    {
                        "first": "Iryna",
                        "middle": [],
                        "last": "Gurevych",
                        "suffix": ""
                    },
                    {
                        "first": "Torsten",
                        "middle": [],
                        "last": "Zesch",
                        "suffix": ""
                    }
                ],
                "year": 2012,
                "venue": "conjunction with the First Joint Conference on Lexical and Computational Semantics",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Daniel B., Chris Biemann, Iryna Gurevych and Torsten Zesch. 2012. UKP: Computing Semantic Textual Similarity by Combining Multiple Content Similarity Measures. Proceedings of the 6th International Workshop on Semantic Evaluation (SemEval 2012), in conjunction with the First Joint Conference on Lexical and Computational Semantics (*SEM 2012).",
                "links": null
            },
            "BIBREF1": {
                "ref_id": "b1",
                "title": "SemEval-2012 Task 6: A Pilot on Semantic Textual Similarity",
                "authors": [
                    {
                        "first": "Eneko",
                        "middle": [],
                        "last": "Agirre",
                        "suffix": ""
                    },
                    {
                        "first": "Daniel",
                        "middle": [],
                        "last": "Cer",
                        "suffix": ""
                    },
                    {
                        "first": "Mona",
                        "middle": [],
                        "last": "Diab",
                        "suffix": ""
                    },
                    {
                        "first": "Aitor",
                        "middle": [],
                        "last": "Gonzalez-Agirre",
                        "suffix": ""
                    }
                ],
                "year": 2012,
                "venue": "conjunction with the First Joint Conference on Lexical and Computational Semantics (*SEM 2012",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Eneko Agirre, Daniel Cer, Mona Diab, and Aitor Gonza- lez-Agirre. 2012. SemEval-2012 Task 6: A Pilot on Semantic Textual Similarity. Proceedings of the 6th International Workshop on Semantic Evaluation (SemEval 2012), in conjunction with the First Joint Conference on Lexical and Computational Semantics (*SEM 2012).",
                "links": null
            },
            "BIBREF2": {
                "ref_id": "b2",
                "title": "TakeLab: Systems for Measuring Semantic Text Similarity",
                "authors": [
                    {
                        "first": "Frane",
                        "middle": [],
                        "last": "Saric",
                        "suffix": ""
                    },
                    {
                        "first": "Goran",
                        "middle": [],
                        "last": "Glavas",
                        "suffix": ""
                    },
                    {
                        "first": "Mladen",
                        "middle": [],
                        "last": "Karan",
                        "suffix": ""
                    },
                    {
                        "first": "Jan",
                        "middle": [],
                        "last": "Snajder",
                        "suffix": ""
                    },
                    {
                        "first": "Bojana Dalbelo",
                        "middle": [],
                        "last": "Basia",
                        "suffix": ""
                    }
                ],
                "year": 2012,
                "venue": "conjunction with the First Joint Conference on Lexical and Computational Semantics",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Frane Saric, Goran Glavas, Mladen Karan, Jan Snajder and Bojana Dalbelo Basia. 2012. TakeLab: Systems for Measuring Semantic Text Similarity. Proceedings of the 6th International Workshop on Semantic Evaluation (SemEval 2012), in conjunction with the First Joint Conference on Lexical and Computational Semantics (*SEM 2012).",
                "links": null
            },
            "BIBREF3": {
                "ref_id": "b3",
                "title": "Lexrank: Graph-based Lexical Centrality as Salience in Text Summarization",
                "authors": [
                    {
                        "first": "Gunes",
                        "middle": [],
                        "last": "Erkan",
                        "suffix": ""
                    },
                    {
                        "first": "Dragomir",
                        "middle": [
                            "R"
                        ],
                        "last": "Radev",
                        "suffix": ""
                    }
                ],
                "year": 2004,
                "venue": "Journal of Artificial Intelligence Research",
                "volume": "22",
                "issue": "",
                "pages": "457--479",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Gunes Erkan and Dragomir R. Radev. 2004. Lexrank: Graph-based Lexical Centrality as Salience in Text Summarization. Journal of Artificial Intelligence Research, 22(2004):457-479.",
                "links": null
            },
            "BIBREF4": {
                "ref_id": "b4",
                "title": "Text Classification using String Kernels",
                "authors": [
                    {
                        "first": "Huma",
                        "middle": [],
                        "last": "Lodhi",
                        "suffix": ""
                    },
                    {
                        "first": "Craig",
                        "middle": [],
                        "last": "Saunders",
                        "suffix": ""
                    },
                    {
                        "first": "John",
                        "middle": [],
                        "last": "Shawe-Taylor",
                        "suffix": ""
                    },
                    {
                        "first": "Nello",
                        "middle": [],
                        "last": "Cristianini",
                        "suffix": ""
                    },
                    {
                        "first": "Chris",
                        "middle": [],
                        "last": "Watkins",
                        "suffix": ""
                    }
                ],
                "year": 2002,
                "venue": "The Journal of Machine Learning Research",
                "volume": "2",
                "issue": "",
                "pages": "419--444",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Huma Lodhi, Craig Saunders, John Shawe-Taylor, Nello Cristianini, and Chris Watkins. 2002. Text Classification using String Kernels. The Journal of Machine Learning Research, 2(2002):419-444.",
                "links": null
            },
            "BIBREF5": {
                "ref_id": "b5",
                "title": "PolyUCOMP: Combining Semantic Vectors with Skip-bigrams for Semantic Textual Similarity",
                "authors": [
                    {
                        "first": "Jian",
                        "middle": [],
                        "last": "Xu",
                        "suffix": ""
                    },
                    {
                        "first": "Qin",
                        "middle": [],
                        "last": "Lu",
                        "suffix": ""
                    },
                    {
                        "first": "Zhengzhong",
                        "middle": [],
                        "last": "Liu",
                        "suffix": ""
                    }
                ],
                "year": 2012,
                "venue": "conjunction with the First Joint Conference on Lexical and Computational Semantics (*SEM 2012",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Jian Xu, Qin Lu and Zhengzhong Liu. 2012. PolyUCOMP: Combining Semantic Vectors with Skip-bigrams for Semantic Textual Similarity. Proceedings of the 6th International Workshop on Semantic Evaluation (SemEval 2012), in conjunction with the First Joint Conference on Lexical and Computational Semantics (*SEM 2012).",
                "links": null
            },
            "BIBREF6": {
                "ref_id": "b6",
                "title": "Transferring Topical Knowledge from Auxiliary Long Text for Short Text Understanding",
                "authors": [
                    {
                        "first": "Ou",
                        "middle": [],
                        "last": "Jin",
                        "suffix": ""
                    },
                    {
                        "first": "Nathan",
                        "middle": [
                            "Nan"
                        ],
                        "last": "Liu",
                        "suffix": ""
                    },
                    {
                        "first": "Yong",
                        "middle": [],
                        "last": "Yu",
                        "suffix": ""
                    },
                    {
                        "first": "Qiang",
                        "middle": [],
                        "last": "Yang",
                        "suffix": ""
                    }
                ],
                "year": 2011,
                "venue": "Proceedings of the 20th ACM Conference on Information and Knowledge Management",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Ou Jin, Nathan Nan Liu, Yong Yu and Qiang Yang 2011. Transferring Topical Knowledge from Auxiliary Long Text for Short Text Understanding. Proceedings of the 20th ACM Conference on Information and Knowledge Management (ACM CIKM 2011).",
                "links": null
            },
            "BIBREF7": {
                "ref_id": "b7",
                "title": "Corpusbased and Knowledge-based Measures of Text Semantic Similarity",
                "authors": [
                    {
                        "first": "Rada",
                        "middle": [],
                        "last": "Mihalcea",
                        "suffix": ""
                    },
                    {
                        "first": "Courtney",
                        "middle": [],
                        "last": "Corley",
                        "suffix": ""
                    }
                ],
                "year": 2006,
                "venue": "Proceeding of the Twenty-First National Conference on Artificial Intelligence and the Eighteenth Innovative Applications of Artificial Intelligence Conference",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Rada Mihalcea and Courtney Corley. 2006. Corpusbased and Knowledge-based Measures of Text Semantic Similarity. Proceeding of the Twenty-First National Conference on Artificial Intelligence and the Eighteenth Innovative Applications of Artificial Intelligence Conference..",
                "links": null
            },
            "BIBREF8": {
                "ref_id": "b8",
                "title": "Extended Gloss Overlaps as a Measure of Semantic Relatedness",
                "authors": [
                    {
                        "first": "Satanjeev",
                        "middle": [],
                        "last": "Banerjee",
                        "suffix": ""
                    },
                    {
                        "first": "Ted",
                        "middle": [],
                        "last": "Pedersen",
                        "suffix": ""
                    }
                ],
                "year": 2003,
                "venue": "Proceedings of the 18th International Joint Conference on Artificial Intelligence",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Satanjeev Banerjee and Ted Pedersen. 2003. Extended Gloss Overlaps as a Measure of Semantic Relatedness. Proceedings of the 18th International Joint Conference on Artificial Intelligence.",
                "links": null
            },
            "BIBREF9": {
                "ref_id": "b9",
                "title": "Detecting Text Similarity over Short Passages: Exploring Linguistic Feature Combinations via Machine Learning. Proceeding of Empirical Methods in natural language processing and Very Large Corpora",
                "authors": [
                    {
                        "first": "Vasileios",
                        "middle": [],
                        "last": "Hatzivassiloglou",
                        "suffix": ""
                    },
                    {
                        "first": "Judith",
                        "middle": [
                            "L"
                        ],
                        "last": "Klavans",
                        "suffix": ""
                    },
                    {
                        "first": "Eleazar",
                        "middle": [],
                        "last": "Eskin",
                        "suffix": ""
                    }
                ],
                "year": 1999,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Vasileios Hatzivassiloglou, Judith L. Klavans , Eleazar Eskin. 1999. Detecting Text Similarity over Short Passages: Exploring Linguistic Feature Combinations via Machine Learning. Proceeding of Empirical Methods in natural language processing and Very Large Corpora.",
                "links": null
            },
            "BIBREF10": {
                "ref_id": "b10",
                "title": "Sentence Similarity Based on Semantic Nets and Corpus Statistics",
                "authors": [
                    {
                        "first": "Yuhua",
                        "middle": [],
                        "last": "Li",
                        "suffix": ""
                    },
                    {
                        "first": "David",
                        "middle": [],
                        "last": "Mclean",
                        "suffix": ""
                    },
                    {
                        "first": "B",
                        "middle": [],
                        "last": "Zuhair",
                        "suffix": ""
                    },
                    {
                        "first": "James",
                        "middle": [
                            "D"
                        ],
                        "last": "O'shea",
                        "suffix": ""
                    },
                    {
                        "first": "Keeley",
                        "middle": [],
                        "last": "Crockett",
                        "suffix": ""
                    }
                ],
                "year": 2006,
                "venue": "IEEE Transactions on Knowledge and Data Engineering",
                "volume": "18",
                "issue": "8",
                "pages": "1138--1149",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Yuhua Li, David Mclean, Zuhair B, James D. O'shea and Keeley Crockett. 2006. Sentence Similarity Based on Semantic Nets and Corpus Statistics. IEEE Transactions on Knowledge and Data Engineering, 18(8):1138-1149.",
                "links": null
            }
        },
        "ref_entries": {
            "FIGREF0": {
                "type_str": "figure",
                "text": ": n-word overlap representation",
                "uris": null,
                "num": null
            },
            "FIGREF1": {
                "type_str": "figure",
                "text": "a b c d and b c a d for example, we will have the matrix as follows,",
                "uris": null,
                "num": null
            },
            "TABREF0": {
                "content": "<table/>",
                "num": null,
                "type_str": "table",
                "html": null,
                "text": "Matrix representation for two sentences"
            },
            "TABREF1": {
                "content": "<table/>",
                "num": null,
                "type_str": "table",
                "html": null,
                "text": "Updated matrix representation for two sentences"
            },
            "TABREF2": {
                "content": "<table><tr><td colspan=\"3\">TEAM headlines OnWN FNWN SMT</td><td>mean rank</td></tr><tr><td>RUN1</td><td>0.5176</td><td colspan=\"2\">0.1517 0.2496 0.2914 0.3284 77</td></tr></table>",
                "num": null,
                "type_str": "table",
                "html": null,
                "text": "Feature mapping for two entities"
            },
            "TABREF3": {
                "content": "<table/>",
                "num": null,
                "type_str": "table",
                "html": null,
                "text": "Experimental results for STS-CORE"
            },
            "TABREF5": {
                "content": "<table><tr><td>Type</td><td>Strategy</td></tr><tr><td>author</td><td>String kernel</td></tr><tr><td>people</td><td>String kernel + Jaccard</td></tr><tr><td>time</td><td>String kernel + Jaccard</td></tr><tr><td>location</td><td>String kernel + Jaccard</td></tr><tr><td>event</td><td>WordNet + Jaccard</td></tr><tr><td>subject</td><td>Sense overlapping</td></tr><tr><td>description</td><td>Sense overlapping</td></tr></table>",
                "num": null,
                "type_str": "table",
                "html": null,
                "text": "Experimental results for STS-TYPED"
            },
            "TABREF6": {
                "content": "<table/>",
                "num": null,
                "type_str": "table",
                "html": null,
                "text": "Strategies for computing similarity"
            }
        }
    }
}