Deep Semantic Parsing with Upper Ontologies
Round 1
Reviewer 1 Report
The paper's aim is to present a new method for parsing various corpora based on well-known vocabularies and conceptual models, in order to obtain semantic roles for a specific kind of applications (here, virtual reality physics engine simulators). In literature, there are many approaches to determine relevant entities with respect to a given task/domain. The article seems to replicate these findings. The goal of generating 3D scenes based on different textual discourses is not fulfilled.
There are many considerations regarding common NLP operations and WordNet parsing. Algorithms' complexity should be proved (lines 356–374). Why not considering a specific physical 3D space-oriented ontology -- for instance, consult https://arxiv.org/pdf/1803.08495.pdf --, instead of using a too general upper level ontology and the k-NN processing the WordNet synsets, FrameNet and related glossaries and thesauri? What were the difficulties on using Drools? What if the text is illogical, so the ontological model is inconsistent? Comparative approaches on a large number of texts?
The goal mentioned in the introduction (i.e. generating 3D scenes from text documents) is not accomplished according to the methodology described and experimented in sections 4 and 5. Processing a text about Sherlock Holmes has nothing to do with 3D physics engine simulators. The paper fails to convince the reader what are the significant aspects of research. Also, it is very hard to follow (it should be entirely restructured).
Several techniques and methods are quite old. Instead of Cyc, the KDpedia <https://kbpedia.org/> could a suitable solution. Also, lemon <https://lemon-model.net/> could be considered. Also, multi-lingual aspects are not mentioned.
All diagrams should be professionally remade. From section 5 (line 477), the paper jumps directly to section 7 (line 548). Related work should be expanded. Several bibliographic references are (very) old.
Author Response
This video explains the functionality of the system and is intended for a wider audience.
Video Link: https://youtu.be/JG3OkZ7_8uw <https://youtu.be/JG3OkZ7_8uw>
Author Response File: 
Author Response.docx
Reviewer 2 Report
The paper deals with a Deep semantic parsing with upper ontologies.
According to the reviewer, the paper is worth publishing at Applied Sciences Journal,
but some corrections are needed and then the paper can be accepted for publication in the journal.
While the authors have made considerable research effort,
the presentation of the paper and the results must be proved.
Additionally make the following corrections to the manuscript:
Comment 1
The authors use a high number of the word "we".
It's not so good to use the word "we".
Comment 2
Figure 2
The authors must improve the visibility of the Figure.
Comment 3
Line 127
What is the (§3).?
Comment 4
Figure 4
The processes in bold indicate
The authors must improve the visibility of the The processes in bold.
Comment 5
Lines 140, 150, 164 and 177
1, 2 and 4: no bold
3: bold
The authors must change the format (the same for all stages).
Comment 6
Figure 3: There are 4 main stages
Line 137: These five off-the-shelf components can run
The authors must explain the difference between stage and component.
Comment 7
Table 1 title must be with the Table on the same page.
Comment 8
The numbering of the equations should be at the end of the line on the right side of the text.
Comment 9
Line 336
Figure 6. Frame identification
Figure 6. (format: bold)
Comment 10
Line 383
Using the Stanford dependency
replace (delete the extra spaces)
Comment 11
Line 387
Word sense disambiguation (WSD)
replace
Word Sense Disambiguation (WSD)
Comment 12
Line 405
The authors must insert a Figure 7 title.
a, b and c: bold
d: no bold
Comment 13
Increase the number of the reference papers including (primarily) from Applied Sciences.
The authors use 0 paper from Appied Sciences journal / 0 papers from MDPI Journals / 32 papers from journals (References)
Τhe number for papers from MDPI journals
is considered insufficient (in reviewer's opinion).
Comment 14
References
The authors must format the References according to the journal's instructions
(References should be described as follows, depending on the type of work:
Journal Articles:
1. Author 1, A.B.; Author 2, C.D. Title of the article. Abbreviated Journal Name Year, Volume, page range.)
Comment 15
The authors must consider a statistical point of view for results presentation.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
Previous Comment 1
The authors use a high number of the word "we".
It's not so good to use the word "we".
Authors' Responses to Reviewer's Comments:
1. We reduced the number of the word "we" that used in previous version of the paper.
We agree that for formal writing excessive use of “we” is not so good.
But, in the revision paper, the number of the word "we":
Line 23
Line 53
Line 140
Line 151
Line 156
Line 161
Line 164
Line 167
Line 183
Line 208
Line 209
Line 226
Line 244
Line 266
Line 272
Line 303
Line 310
Line 322
Line 325
Line 326
Line 340
Line 347
Line 349
Line 359
Line 361
Line 371
Line 372
Line 380
Line 383
Line 406 two times
Line 416
Line 418
Line 421
Line 423
Line 425
Line 433
Line 453 two times
Line 454
Line 459
Line 462
Line 466
Line 467
Line 468
Line 470
Line 474
Line 480
Line 483
Line 488
Line 490
Line 494
Line 495
Line 502
Line 505
Line 508
Line 511
Line 513
Line 516
Line 517
Line 524 two times
Line 525
Line 526
Line 527
Line 528 two times
Line 541
Line 542
Line 547
Line 556
Line 560
Line 561
Line 569
Line 570
Line 575
Line 576
Line 578
Line 579
Line 582
Line 583
Line 586
Line 590
Line 591
Line 593
Line 595
Line 602
Line 609
Line 614
Line 615
Line 617
Line 626
Line 630
Line 631
Line 638
Line 642
Line 644
Line 649
Line 650 two times
Line 651
Line 657
Line 660
Line 663 two times
Line 664
Previous Comment 14
References
The authors must format the References according to the journal's instructions
(References should be described as follows, depending on the type of work:
Journal Articles:
1. Author 1, A.B.; Author 2, C.D. Title of the article. Abbreviated Journal Name Year, Volume, page range.)
Authors' Responses to Reviewer's Comments:
We have made changes in the References list according to the instructions of the journal.
References Format is not according to the journal's instructions.
Author Response
Comment 1
We reduced the number of the word "we" that used in previous version of the paper from 100+ to 7. This is below the average number of papers usually accepted in the Applied Sciences Journal.
Comment 14
We have made changes in the References list according to the instructions of the journal.
We've checked each article in the References list and added additional information and formatted the links accordingly.
Round 3
Reviewer 2 Report
Lines 357 and 434
Figures 6 and 7.
The authors must format the text (bold).
Author Response
We've changed the fonts in "Figure 6." and "Figure 7."
