File size: 37,898 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
{
    "paper_id": "W07-0209",
    "header": {
        "generated_with": "S2ORC 1.0.0",
        "date_generated": "2023-01-19T04:40:10.924097Z"
    },
    "title": "Semi-supervised Algorithm for Human-Computer Dialogue Mining",
    "authors": [
        {
            "first": "Calkin",
            "middle": [
                "S"
            ],
            "last": "Montero",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "Hokkaido University",
                "location": {
                    "addrLine": "Kita 14-jo Nishi 9-chome Kita-ku",
                    "postCode": "060-0814",
                    "settlement": "Sapporo",
                    "country": "Japan"
                }
            },
            "email": "[email protected]"
        },
        {
            "first": "Kenji",
            "middle": [],
            "last": "Araki",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "Hokkaido University",
                "location": {
                    "addrLine": "Kita 14-jo Nishi 9-chome Kita-ku",
                    "postCode": "060-0814",
                    "settlement": "Sapporo",
                    "country": "Japan"
                }
            },
            "email": "[email protected]"
        }
    ],
    "year": "",
    "venue": null,
    "identifiers": {},
    "abstract": "This paper describes the analysis of weak local coherence utterances during humancomputer conversation through the application of an emergent data mining technique, data crystallization. Results reveal that by adding utterances with weak local relevance the performance of a baseline conversational partner, in terms of user satisfaction, showed betterment.",
    "pdf_parse": {
        "paper_id": "W07-0209",
        "_pdf_hash": "",
        "abstract": [
            {
                "text": "This paper describes the analysis of weak local coherence utterances during humancomputer conversation through the application of an emergent data mining technique, data crystallization. Results reveal that by adding utterances with weak local relevance the performance of a baseline conversational partner, in terms of user satisfaction, showed betterment.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Abstract",
                "sec_num": null
            }
        ],
        "body_text": [
            {
                "text": "Data mining can be defined as the process of finding new and potentially useful knowledge from data. An enhanced trend of data mining is chance discovery, which in spite of being an emergent field of research has been applied to different branches of science. Recently, data crystallization (Ohsawa, 2005) has been proposed as a chance discovery extension devoted to find desired unobservable events within a given data. This method of data mining has been intended for revealing events that are significant but are not included in the analyzed data (sparseness).",
                "cite_spans": [
                    {
                        "start": 291,
                        "end": 305,
                        "text": "(Ohsawa, 2005)",
                        "ref_id": "BIBREF12"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Introduction",
                "sec_num": "1"
            },
            {
                "text": "The knowledge database of a conversational partner computer program is a good example of data sparseness, making very difficult to realize a humanlike conversation between a user and a computer, since the database is missing relevant or coherent answers for a given user input, e.g., ELIZA system (Weizenbaum, 1966) , corpus-based dialogue systems (Inui et al., 2003) , and so forth. Herring (1999) noticed the weak sequential coherence during human-computer interaction that is caused by weak local relevance, pointing out the violation of the Gricean maxim of local relevance during the dialogue. However, we argue that conversational responses -often of very weak local relevance -generated by a conversational partner computer program (or \"chatbot\") could successfully simulate humanlike coherent dialogue (see example given by (Herring, 1999) , where topical coherence is stated in spite of violation of local relevance Gricean maxim). Moreover, we argue that weak local relevance utterances could be regarded as topic-shifting utterances. These utterances are vital in order to keep the flow of the dialogue and its overall coherence. Determining the moment when the topic-shifting utterances are needed is hence essential to the success of a chatbot.",
                "cite_spans": [
                    {
                        "start": 297,
                        "end": 315,
                        "text": "(Weizenbaum, 1966)",
                        "ref_id": "BIBREF15"
                    },
                    {
                        "start": 348,
                        "end": 367,
                        "text": "(Inui et al., 2003)",
                        "ref_id": "BIBREF11"
                    },
                    {
                        "start": 384,
                        "end": 398,
                        "text": "Herring (1999)",
                        "ref_id": "BIBREF10"
                    },
                    {
                        "start": 832,
                        "end": 847,
                        "text": "(Herring, 1999)",
                        "ref_id": "BIBREF10"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Introduction",
                "sec_num": "1"
            },
            {
                "text": "In this paper KeyGraph and data crystallization techniques are applied to analyze human-computer trivial dialogue, chat, considering as chances to discover the missing topic-shifting utterances, used to improve the overall sequential coherence and the flow of the computer dialogue.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Introduction",
                "sec_num": "1"
            },
            {
                "text": "During the process of dialogue mining, in order to analyze the behavior of human-computer conversational interaction, the interrelationship of utterances of a dialogue section was obtained. The following is a fragment of an analyzed dialogue section between a user and ALICE chatbot (Wallace, 2005 For the analysis, the application of data crystallization is introduced in order to identify the missing topic-shifting utterances. The interrelationship of utterances (user-chatbot) is represented in a cooccurrence document created as described below.",
                "cite_spans": [
                    {
                        "start": 283,
                        "end": 297,
                        "text": "(Wallace, 2005",
                        "ref_id": "BIBREF14"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Human-Computer Conversation Analysis Approach",
                "sec_num": "2"
            },
            {
                "text": "The KeyGraph has been used as a data-mining tool for extracting patterns of the appearance of chance events (Ohsawa et al.(2003) ). The KeyGraph identifies relationships between terms in a document particularly focusing on co-occurrence relationships of both high-probability and low-probability events. Montero et al. (2005) have applied this tool for analyzing the dynamic behavior of human-human chat, identifying criticality.",
                "cite_spans": [
                    {
                        "start": 108,
                        "end": 128,
                        "text": "(Ohsawa et al.(2003)",
                        "ref_id": null
                    },
                    {
                        "start": 319,
                        "end": 325,
                        "text": "(2005)",
                        "ref_id": null
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "The KeyGraph",
                "sec_num": "2.1"
            },
            {
                "text": "In this paper the KeyGraph is applied in combination with data crystallization in order to visualized utterances that do not appear during humancomputer chat. The interrelationship of utterances (user-chatbot) is represented in a co-occurrence document created by the following algorithm: a) Each utterance (from both, the user and the chatbot) was considered as Then it could be said that the obtained D document contains the co-occurrence relationship of the utterances during the analyzed dialogue section. In the graph, the most frequent items in D are shown as black nodes and the most strongly co-occurring item-pairs are linked by black lines according to the Jaccard coefficient:",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "The KeyGraph",
                "sec_num": "2.1"
            },
            {
                "text": "J(S x , S y ) = p(S x \u2229 S y )/p(S x \u222a S y )",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "The KeyGraph",
                "sec_num": "2.1"
            },
            {
                "text": "where p(S x \u2229 S y ) is the probability that both elements S x and S y co-occur in a line in D, and p(S x \u222a S y ) is the probability that either S x or S y appears in a line. In the graph, nodes are interpreted as sentences (from D) and clusters of nodes as particular topics (Figure 1 ).",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 275,
                        "end": 284,
                        "text": "(Figure 1",
                        "ref_id": "FIGREF2"
                    }
                ],
                "eq_spans": [],
                "section": "The KeyGraph",
                "sec_num": "2.1"
            },
            {
                "text": "Data crystallization (Ohsawa, 2005) , is dedicated to experts working in real domains where discoveries of events that are important but are not included in the analyzed data are desired. The process of data crystallization involves to insert dummy items in the given data in order to represent unobservable events. In this paper, each dummy item inserted in the D document (one in each vector of D) is named X Y , where X represents the level of the insertion and Y represents the line where the dummy item was inserted. The KeyGraph is applied to the new D document and all of the dummy nodes that did not appear linking clusters in the graph are eliminated from the data, and then the cycle is iterated to higher levels. In the case of the D document of Sec.2.1, after the first level of insertion it becomes: After feeding the KeyGraph with D \u2032 , all the dummy items that did not appear linking clusters as bridges in the outputted graph are deleted. At this point new dummy items with higher hierarchy (2 x ) are inserted in D \u2032 , and the cycle iterates. Unobservable events and their relations with other events are to be visualized by the application of KeyGraph iteratively to the data that is been crystallized (Figure 2 ). ",
                "cite_spans": [
                    {
                        "start": 21,
                        "end": 35,
                        "text": "(Ohsawa, 2005)",
                        "ref_id": "BIBREF12"
                    }
                ],
                "ref_spans": [
                    {
                        "start": 1220,
                        "end": 1229,
                        "text": "(Figure 2",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Data Crystallization",
                "sec_num": "2.2"
            },
            {
                "text": "D \u2032 =",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Data Crystallization",
                "sec_num": "2.2"
            },
            {
                "text": "The performed experiment was carried out in three stages. In the first stage of the experiment, three different dialogue sections (including the one shown in Sec.2) between three native English speakers and a chatbot (Wallace, 2005) were analyzed in order to find co-occurrence between the users' utterance and the chatbot replies, i.e., D document. This D document was then examined by the KeyGraph (unsupervised process). Figure 1 shows the graphical view of the dialogue in Sec.2 (48 turns, user -chatbot, in total). A characteristic of the KeyGraph is the visualization of co-occurring events by means of clusters. In Figure 1 , the nodes represent sentences from the D document, the clusters represent the relationship among those sentences, i.e., a specific topic, and the nodes that link the clusters represent the transition from one topic to the next. It can be observed that the main clusters are not interconnected, leading to the conclusion that the chatbot in many cases could not keep a smooth and natural flow of the dialogue.",
                "cite_spans": [
                    {
                        "start": 217,
                        "end": 232,
                        "text": "(Wallace, 2005)",
                        "ref_id": "BIBREF14"
                    }
                ],
                "ref_spans": [
                    {
                        "start": 424,
                        "end": 432,
                        "text": "Figure 1",
                        "ref_id": "FIGREF2"
                    },
                    {
                        "start": 622,
                        "end": 630,
                        "text": "Figure 1",
                        "ref_id": "FIGREF2"
                    }
                ],
                "eq_spans": [],
                "section": "Experiment and Visual Results",
                "sec_num": "3"
            },
            {
                "text": "In the second stage of the experiment, a crystallized document of utterance co-occurrence, i.e., D \u2032 document, was obtained for the same dialogue sections, following the process described in Sec.2.2. The graphical output of the dialogue in Sec.2, after crystallization, can be observed in Figure 2 . It can be seen in this figure how the two main clusters appear to be interconnected by the dummy item 1 3.",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 289,
                        "end": 297,
                        "text": "Figure 2",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Experiment and Visual Results",
                "sec_num": "3"
            },
            {
                "text": "Although this dummy item was inserted in the third line of the D document, it appears in the graph connecting the two main clusters. The dummy item 1 3 branches from utterance [24] . This interconnecting point can be regarded as the system considering it appropriate to insert a topic-shifting utterance at this point of the conversation. In doing so, a well interconnected graph is obtained (Figure 2 ). This information is valuable for making the chatbot to ask \"intelligent questions\" as a mean of conversational responses to keep the interest from the user.",
                "cite_spans": [
                    {
                        "start": 176,
                        "end": 180,
                        "text": "[24]",
                        "ref_id": "BIBREF9"
                    }
                ],
                "ref_spans": [
                    {
                        "start": 392,
                        "end": 401,
                        "text": "(Figure 2",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Experiment and Visual Results",
                "sec_num": "3"
            },
            {
                "text": "In the third stage of the experiment, the information yielded by the previous analysis, i.e., regarding the timing where a topic-shifting utterance might be needed, was used to feed the chatbot database. Topic-shifting responses were inserted by hand (supervised process) as general patterns (around one hundred patterns) for smoothly change the topic when there is not a pattern that matches a given utterance. In this way a bridge, represented in Figure  2 by the dummy item, is created giving to the dialogue the desired smoothness. Seven users (four native English speakers, three non native speakers) were requested to perform a chat with the plain chatbot and with the enhanced chatbot (the users did not know which chatbot was plain or which was enhanced). The time set up was maximum 30 minuteschat with each program, the user was free to stop at any time before the time limit. The evaluation of the chatbots performances was made through a ques- The four native English speaker users globally agreed ranking enhanced chatbot as having \"good\" accurate/relevant responses during the overall dialogue, giving remarks like \"interesting to talk with\", \"better at keeping the conversation going\", \"easier to talk to\", and \"more impression of consciousness\". In the case of the plain chatbot, the native speakers showed \"dissatisfied\" at its performance, giving remarks like \"loop responses\", \"slow in keeping the conversation going\", \"no so nice responses\", \"no constant memory\". Table 1 shows a resume of the average performance of the chatbot for all of the users. An example of a vague reply is given in the following fragment: Two non native English speaker users ranked the enhanced chatbot as having \"fair\" and \"average\" accurate/relevant responses while the plain chatbot was ranked as having \"poor\" and \"fair\" accurate/relevant responses. The third non native English speaker user ranked both chatbots as \"poor\" due to \"the chatbots lack of understanding deixis, and anaphor\".",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 449,
                        "end": 458,
                        "text": "Figure  2",
                        "ref_id": null
                    },
                    {
                        "start": 1484,
                        "end": 1491,
                        "text": "Table 1",
                        "ref_id": "TABREF2"
                    }
                ],
                "eq_spans": [],
                "section": "Experiment and Visual Results",
                "sec_num": "3"
            },
            {
                "text": "As a mean of discussion, in Figure 2 it could be expected that the dummy item 1 3 would branch from utterance [25] {User: no, you asked me who is the best robot}, which is in the same cluster with utterance [24] . However, under closer examination it becomes clear that utterance [24] has stronger co-occurrence with utterance [38] {Chatbot: I know you are but what am I} than utterance [25] . Hence, the algorithm suggests to link the clusters via utterance [24] .",
                "cite_spans": [
                    {
                        "start": 207,
                        "end": 211,
                        "text": "[24]",
                        "ref_id": "BIBREF9"
                    },
                    {
                        "start": 280,
                        "end": 284,
                        "text": "[24]",
                        "ref_id": "BIBREF9"
                    },
                    {
                        "start": 387,
                        "end": 391,
                        "text": "[25]",
                        "ref_id": null
                    },
                    {
                        "start": 459,
                        "end": 463,
                        "text": "[24]",
                        "ref_id": "BIBREF9"
                    }
                ],
                "ref_spans": [
                    {
                        "start": 28,
                        "end": 36,
                        "text": "Figure 2",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Experiment and Visual Results",
                "sec_num": "3"
            },
            {
                "text": "In other aspect, based on the feedback given by the seven users of the experiment, the overall performance of the enhanced chatbot can be considered better than the plain chatbot. It is worth noticing that the evaluation of the non native English speaker users tended to emphasize the grammatical aspect of the chatbots responses. On the other hand, the evaluation of the native English speaker users tended to emphasize the smoothness of the dialogue. Although there is still plenty of room for improvement and research a betterment in the chatbot performance could be seen through this approach.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Experiment and Visual Results",
                "sec_num": "3"
            },
            {
                "text": "In this paper the application of a novel data mining method, data crystallization, for visualizing missing topic-shifting utterances during human-computer chat has been described. Based on this information, during the experiment, the use of weak local relevance utterances, i.e., topic-shifting responses, despite of violation of Grecian maxim of local relevance, showed to meliorate the overall dialogue flow. Future research will be oriented to the extended implementation of the obtained results for enhancing the chat flow modeling of a conversational partner program.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Conclusion",
                "sec_num": "4"
            },
            {
                "text": "Since follow-ups were eliminated, the number of sentences in D might be smaller than the actual number of sentences in the dialogue.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            }
        ],
        "back_matter": [],
        "bib_entries": {
            "BIBREF0": {
                "ref_id": "b0",
                "title": "User: what should I call you?",
                "authors": [],
                "year": null,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "User: what should I call you?",
                "links": null
            },
            "BIBREF1": {
                "ref_id": "b1",
                "title": "Chatbot: my name is Alice",
                "authors": [],
                "year": null,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Chatbot: my name is Alice",
                "links": null
            },
            "BIBREF2": {
                "ref_id": "b2",
                "title": "User: what do you mean by robot? there are lots of different robots",
                "authors": [],
                "year": null,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "User: what do you mean by robot? there are lots of differ- ent robots",
                "links": null
            },
            "BIBREF3": {
                "ref_id": "b3",
                "title": "Have you tried a web search? Would you like to know more?",
                "authors": [
                    {
                        "first": "",
                        "middle": [],
                        "last": "Chatbot",
                        "suffix": ""
                    }
                ],
                "year": null,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Chatbot: It is a good question. Have you tried a web search? Would you like to know more?",
                "links": null
            },
            "BIBREF4": {
                "ref_id": "b4",
                "title": "User: sure",
                "authors": [],
                "year": null,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "User: sure",
                "links": null
            },
            "BIBREF5": {
                "ref_id": "b5",
                "title": "Chatbot: Ok let's do it",
                "authors": [],
                "year": null,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Chatbot: Ok let's do it. What would you like to know?",
                "links": null
            },
            "BIBREF6": {
                "ref_id": "b6",
                "title": "User: what Alice thinks a robot means",
                "authors": [],
                "year": null,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "User: what Alice thinks a robot means",
                "links": null
            },
            "BIBREF7": {
                "ref_id": "b7",
                "title": "Chatbot: is that a rhetorical question? I would search for it",
                "authors": [],
                "year": null,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Chatbot: is that a rhetorical question? I would search for it",
                "links": null
            },
            "BIBREF8": {
                "ref_id": "b8",
                "title": "User: when I hear robot, I think of something that moves",
                "authors": [],
                "year": null,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "User: when I hear robot, I think of something that moves",
                "links": null
            },
            "BIBREF9": {
                "ref_id": "b9",
                "title": "Chatbot: you never know what to expect from life. Why? oh yeah, we are talking about something that moves",
                "authors": [],
                "year": null,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Chatbot: you never know what to expect from life. Why? oh yeah, we are talking about something that moves [....] References",
                "links": null
            },
            "BIBREF10": {
                "ref_id": "b10",
                "title": "Interactional coherence in cmc",
                "authors": [
                    {
                        "first": "Susan",
                        "middle": [],
                        "last": "Herring",
                        "suffix": ""
                    }
                ],
                "year": 1999,
                "venue": "Journal of Computer Mediated Communication",
                "volume": "4",
                "issue": "1",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Susan Herring. 1999. Interactional coherence in cmc. Journal of Computer Mediated Communication, 4(1).",
                "links": null
            },
            "BIBREF11": {
                "ref_id": "b11",
                "title": "Fully corpus-based natural language dialogue system",
                "authors": [
                    {
                        "first": "Nobuo",
                        "middle": [],
                        "last": "Inui",
                        "suffix": ""
                    },
                    {
                        "first": "Takuya",
                        "middle": [],
                        "last": "Koiso",
                        "suffix": ""
                    },
                    {
                        "first": "Junpei",
                        "middle": [],
                        "last": "Nakamura",
                        "suffix": ""
                    },
                    {
                        "first": "Yoshiyuki",
                        "middle": [],
                        "last": "Kotani",
                        "suffix": ""
                    }
                ],
                "year": 2003,
                "venue": "Natural Language Generation in Spoken and Written Dialogue, AAAI Spring Symposium",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Nobuo Inui, Takuya Koiso, Junpei Nakamura, and Yoshiyuki Kotani. 2003. Fully corpus-based natural language dia- logue system. In Natural Language Generation in Spoken and Written Dialogue, AAAI Spring Symposium.",
                "links": null
            },
            "BIBREF12": {
                "ref_id": "b12",
                "title": "Data crystallization: Chance discovery extended for dealing with unobservable events",
                "authors": [
                    {
                        "first": "Yukio",
                        "middle": [],
                        "last": "Ohsawa",
                        "suffix": ""
                    }
                ],
                "year": 2005,
                "venue": "New Mathematics and Natural Computation",
                "volume": "1",
                "issue": "3",
                "pages": "373--392",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Yukio Ohsawa. 2005. Data crystallization: Chance discovery extended for dealing with unobservable events. New Mathe- matics and Natural Computation, 1(3):373-392.",
                "links": null
            },
            "BIBREF13": {
                "ref_id": "b13",
                "title": "Human chat and selforganized criticality: A chance discovery application",
                "authors": [
                    {
                        "first": "S",
                        "middle": [],
                        "last": "Calkin",
                        "suffix": ""
                    },
                    {
                        "first": "Kenji",
                        "middle": [],
                        "last": "Montero",
                        "suffix": ""
                    },
                    {
                        "first": "",
                        "middle": [],
                        "last": "Araki",
                        "suffix": ""
                    }
                ],
                "year": 2005,
                "venue": "New Mathematics and Natural Computation",
                "volume": "1",
                "issue": "3",
                "pages": "407--420",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Calkin S Montero and Kenji Araki. 2005. Human chat and self- organized criticality: A chance discovery application. New Mathematics and Natural Computation, 1(3):407-420.",
                "links": null
            },
            "BIBREF14": {
                "ref_id": "b14",
                "title": "A.l.i.c.e. artificial intelligence foundation",
                "authors": [
                    {
                        "first": "Richard",
                        "middle": [],
                        "last": "Wallace",
                        "suffix": ""
                    }
                ],
                "year": 2005,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Richard Wallace. 2005. A.l.i.c.e. artificial intelligence founda- tion. http://www.alicebot.org.",
                "links": null
            },
            "BIBREF15": {
                "ref_id": "b15",
                "title": "Eliza a computer program for the study of natural language communication between man and machine",
                "authors": [
                    {
                        "first": "Joseph",
                        "middle": [],
                        "last": "Weizenbaum",
                        "suffix": ""
                    }
                ],
                "year": 1966,
                "venue": "Commun. ACM",
                "volume": "9",
                "issue": "1",
                "pages": "36--45",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Joseph Weizenbaum. 1966. Eliza a computer program for the study of natural language communication between man and machine. Commun. ACM, 9(1):36-45.",
                "links": null
            }
        },
        "ref_entries": {
            "FIGREF0": {
                "type_str": "figure",
                "num": null,
                "uris": null,
                "text": "one sentence. b) Each sentence was segmented into words. c) High frequency words were eliminated, i.e., I, you, is, followups and the like, as to avoid false co-occurrence. d) A vectorial representation of each sentence (at word level) was obtained and sentences co-occurrence relationship was determined as 1 : D= w1:: S1, S2, S4 .../ w2:: S9, S25 .../ w3:: S1, S3, S10 .../ ... / wn:: S24, S25, ... Smwhere: w k (k = 1, 2, 3, ..., n), represents a word in a sentence. S l (l = 1, 2, 3, ..., m), represents a sentence."
            },
            "FIGREF1": {
                "type_str": "figure",
                "num": null,
                "uris": null,
                "text": "w1:: S1, S2, S4 ...11 / w2:: S9, S25 ...12 / w3:: S1, S3, S10 ...13 / ... / wn:: S24, S25, ... Sm 1n where 1 o (o = 1, 2, 3, ..., n), represents each dummy item inserted in each vector of D."
            },
            "FIGREF2": {
                "type_str": "figure",
                "num": null,
                "uris": null,
                "text": "User-Computer Chat Graph Figure 2: Crystallized Data Graph"
            },
            "FIGREF3": {
                "type_str": "figure",
                "num": null,
                "uris": null,
                "text": "What is like living in Sapporo? User: sweet as Chatbot: In the context of living in Sapporo I don't understand \"what is like living in Sapporo?\" [vague reply]"
            },
            "TABREF1": {
                "type_str": "table",
                "html": null,
                "num": null,
                "text": ".R.* = Vague Reply (This table contains the average of Turns and VR)",
                "content": "<table><tr><td/><td colspan=\"2\">Turns % V.R.*</td><td>Accuracy/relevance</td></tr><tr><td>Baseline</td><td>74</td><td>21.11%</td><td>fair</td></tr><tr><td>Enhanced</td><td>128</td><td>7.37%</td><td>good</td></tr></table>"
            },
            "TABREF2": {
                "type_str": "table",
                "html": null,
                "num": null,
                "text": "Chatbots Overall Performance tionnaire given to the users. In the questionnaire the degree of satisfaction of the user was surveyed based on the accuracy/relevance of the chatbot responses and its ability to keep the dialogue flowing.",
                "content": "<table><tr><td>The ac-</td></tr></table>"
            }
        }
    }
}