/media/bill/PROJECTS/2019 IJCNN Budapest/Publications/CrossCheck/CrossCheck - example paper comments 20301-20520.txt
www.BillHowell.ca  19Jan2019 initial

This document serves as a detailed proof of CrossCheck analysis of selected IJCNN2019 papers.


********************************************
QUICK SUMMARY OF HOWELL'S ASSESSMENTS 
overall similarities given as (example) "59.0% (75%)" - first is CrossCheck, second is my own eyeball guess
My estimates should generally be lower as I don't include the references, but then I don't include sapce of images.

REJECT as serious attempt to republish a paper or a HOAX!!!?  (possible author bans? but I don't think these are authors)
   N-20471.pdf 98%,  REJECT as serious attempt to republish a paper or a HOAX!!!? : 20471  123558@qq.com 
   N-20442.pdf 96%,  REJECT as serious attempt to republish a paper or a HOAX!!!? Check ALL ! : Serban Iulian [20442 serban@gmail.com, 19973 allen@gmail.com]
   N-20466.pdf, N-20468.pdf 	96%, REJECT BOTH as serious attempts to republish [AAAI-16. Elsevier] papers or HOAX!!!? Check ALL ! : 20466, 20468 nicolasparisi90@gmail.com, 
   N-20306.pdf 	94.0%,  REJECT as serious attempt to republish a paper or a HOAX!!!? : enrique.fan34@gmail.com 

Reject as similarities too high,  
   N-20473.pdf 	59.0% (75%), REJECT -  similarities are from non-authors, mostly short phrase-level similarity with one paragraph, but a few most-paragraph copies
   N-20406.pdf 	52.0% (50%), Reject unless reword or cite own previous work,  Previous author papers account for essentially all similarity, paragraphs & sentences, other authors <=3% similarity
   N-20331.pdf 	87.0%,  Reject - almost all text is in paragraph chunks from other uncited papers, 49% same authors, >49% self-cited, non-author similarities too

Reject arXiv-like postings that are not currently allowed by the IEEE :
   N-20417.pdf 	58.0% (39%), Reject?  OK only if biorxiv.org is accepted by IEEE!! biorxiv.org pre-posted, other diverse authors <=5% similarity combined

Normal reject, but possible OK?
   N-20386.pdf 	52.0% (45%), Reject? Borderline case - my estimate of similarity is in acceptable range, also OK if high self-plagiarism in Introduction acceptable, >70% (?) of similarity is from authors' previous paper, paragraphs in Intro mostly, sentences elsewhere
   N-20478.pdf 	51.0% (37%), Reject? Borderline case - my estimate of similarity is in acceptable range, most similarities are from other authors

OK - arXiv pre-publication by same authors
   N-20486.pdf 	86.0%, OK -  arXiv paper by same authors
   N-20337.pdf 	83.0%, OK -  arXiv paper by same author, other similarities <=4% but add <= 1% to overall similarity
   N-20382.pdf 	82.0%, OK -  arXiv paper by same authors, other similarities <=3% but add <= 2% to overall
   N-20318.pdf 	57.0%, OK -  arXiv paper by same authors, other sources <=3% similarity

Time required to check for papers with >= 50%  similarity :
18Jan2019	8 analysis in 1.8 hours ~= 15 minutes per analysis, (versus 15-30 minutes including past analysis)
				2 batches  in 2.0 hours ~=  1 hour    per batch of 100 papers for [zip download, upload to CrossCheck] 

To generate the list above :
$ grep "^>>>>> Howell :  N-" "/media/bill/PROJECTS/2019 IJCNN Budapest/Publications/CrossCheck/CrossCheck - Howell paper comments 20301-20520.txt"  |  sed  's/>>>>> Howell :  /   /g'
Then sort list into [problems, to check, OK]


********************************************
Howell's detailed assessments 


********************
N-20471.pdf 	98
<.IJCNN paper authors' names.>
<.IJCNN paper title.>

>>>>> Howell :  N-20471.pdf 98%,  REJECT as serious attempt to republish a paper or a HOAX!!!? : 20471  123558@qq.com 

********************
N-20442.pdf 	96
<.IJCNN paper authors' names.>
<.IJCNN paper title.>

>>>>> Howell :  N-20442.pdf 96%,  REJECT as serious attempt to republish a paper or a HOAX!!!? Check ALL ! : Serban Iulian [20442 serban@gmail.com, 19973 allen@gmail.com]

********************
N-20466.pdf 	96%
<.IJCNN paper authors' names.>
<.IJCNN paper title.>

N-20468.pdf 	96%
Parisi Nicolas and Nicolas Parisi 	Deep Neural Adaptation of Sparsity in Multimodal Observations 

>>>>> Howell :  N-20466.pdf, N-20468.pdf 	96%, REJECT BOTH as serious attempts to republish [AAAI-16. Elsevier] papers or HOAX!!!? Check ALL ! : 20466, 20468 nicolasparisi90@gmail.com, 

********************
N-20306.pdf 	94.0
<.IJCNN paper authors' names.>
<.IJCNN paper title.>

>>>>> Howell :  N-20306.pdf 	94.0%,  REJECT as serious attempt to republish a paper or a HOAX!!!? : enrique.fan34@gmail.com 

********************
N-20331.pdf 	87.0
<.IJCNN paper authors' names.>
<.IJCNN paper title.>

Overall view : Previous author paper accounts for most of similarity, phrases mostly, rare sentences
strong similarity : [ALL]

49%	~60 sources >=3% similarity, mostly authors and many <3%
49%	https://ieeexplore.ieee.org/document/7966046
<.Other paper authors' names.>
<.Other paper title.>
>> same authors

24%	https://link.springer.com/book/10.1007%2F978-3-319-26535-3
<.Other paper authors' names.>
<.Other paper title.>
>> same authors

23%	https://link.springer.com/chapter/10.1007%2F978-3-319-26535-3_25
>> same authors

More from same authors :  23, 23, 13, 
13%	from NIPS 2011 paper uncertain non-author

8%		from 12 sources >= 2% similarity, many many more <2%, mix of authors and non-authors

many other groups of <6% similarity (not investigated)

>>>>> Howell :  N-20331.pdf 	87.0%,  Reject - almost all text is in paragraph chunks from other uncited papers, 49% same authors, >49% self-cited, non-author similarities too

********************
N-20486.pdf 	86.0%, 
<.IJCNN paper authors' names.>
<.IJCNN paper title.>

83%	http://export.arxiv.org/pdf/1901.00525
<.Other paper authors' names.>
<.Other paper title.>

No other high similarity sources

>>>>> Howell :  N-20486.pdf 	86.0%, OK -  arXiv paper by same authors

********************
N-20337.pdf 	83.0%
<.IJCNN paper authors' names.>
<.IJCNN paper title.>

83%	11 sources >= 4% similarity, many others <= 1%

82%	http://export.arxiv.org/pdf/1812.1131
Performance of Three Slim Variants of
The Long Short-Term Memory (LSTM) Layer Daniel Kent and Fathi Salem

5%		http://export.arxiv.org/pdf/1901.00525
<.Other paper authors' names.>
<.Other paper title.>

4%		http://export.arxiv.org/list/cs.LG/new
>> different but unknown authors

4%		http://export.arxiv.org/list/cs.AI/new
>> different but unknown authors

4%		http://export.arxiv.org/list/cs/new
>> different but unknown authors

4%		http://export.arxiv.org/list/cs.NE/new

maybe 30 other similarities <1% 

>>>>> Howell :  N-20337.pdf 	83.0%, OK -  arXiv paper by same author, other similarities <=4% but add <= 1% to overall similarity

********************
N-20382.pdf 	82.0%, 
<.IJCNN paper authors' names.>
<.IJCNN paper title.>

+-----+
80%	15 sources, 1 @ 80%, 3 @ 3%, 11 @ 2%

80%	http://export.arxiv.org/pdf/1901.01015
<.Other paper authors' names.>
<.Other paper title.>
>> same authors

>>>>> Howell :  N-20382.pdf 	82.0%, OK -  arXiv paper by same authors, other similarities <=3% but add <= 2% to overall


********************
18Jan2019 16:39 stop here ...
...resume with CrossCheck web-page 2

********************
N-20473.pdf 	59.0%, 
<.IJCNN paper authors' names.>
<.IJCNN paper title.>

Overall view : Mostly short phrase-level similarity with one paragraph, but a few most-paragraph copies
strong similarity (>=50%) : [none ]
moderate (30% < similarity <50%) : [none ] 
low (<30%) :  [ALL]
My eyeball estimate of overall text similarity (not including References) 
= sum(%similar*pagelength) / sum(pagelength) :  
= / [100 * sum EACHBOTH *, sum second] (0.4  1  0.85  0.85  0.4  0) (0.6  0.8  0.6  1  0.4  0) 
= 75 

+-----+
18%	>40 sources, 1 @ 16%, rest <= 1%

16%	https://arxiv.org/pdf/1707.07998.pdf
<.Other paper authors' names.>
<.Other paper title.>
>> one common author

+-----+
10%	~30 sources, 5 of (2% <= similarity <= 10%), rest <= 1%

10%	http://docplayer.net/34831335-Arxiv-v5-cs-cl-10-jun-2016.html
<.Other paper authors' names.>
<.Other paper title.>
>> different authors

7%		4% same as 10% above

9%		https://www.semanticscholar.org/paper/Neural-Machine-Translation-of-Rare-Words-with-Units-Sennrich-Haddow/1518039b5001f1836565215eb047526b3ac7f462
<.Other paper authors' names.>
<.Other paper title.>
>> different authors - same source as above!

9%		same as above
<.Other paper authors' names.>
<.Other paper title.>

+-----+
7%		https://www.mdpi.com/2076-3417/8/5/739/htm
<.Other paper authors' names.>
<.Other paper title.>
>> different authors

>>>>> Howell :  N-20473.pdf 	59.0% (75%), REJECT -  similarities are from non-authors, mostly short phrase-level similarity with one paragraph, but a few most-paragraph copies

********************
N-20417.pdf 	58.0%, 
<.IJCNN paper authors' names.>
<.IJCNN paper title.>

Overall view : Previous author paper accounts almost all similarity, mostly paragraphs
strong similarity (>=50%) : [Abstract, Introduction, Discussion References, ]
moderate (30% < similarity <50%) : [Material & method BCG, Research Status, ] 
low (<30%) :  [Material & method ADEF, Results,]
My eyeball estimate of overall text similarity (not including References) 
= sum(%similar*pagelength) / sum(pagelength) :  
= / [100 * sum EACHBOTH *, sum second]  (0.2  0.75  0.05  0.25  0.35  0.6  0.75  1) (0.6  0.8  1  0.65  0.5  0.75  0.3  0) 
= 38.75 

33%	1 source mostly, others <=1%

33%	https://www.biorxiv.org/content/early/2018/11/24/475855
<.Other paper authors' names.>
<.Other paper title.>
>> 3/4 same authors

All other source groups <= 1%

>>>>> Howell :  N-20417.pdf 	58.0% (39%), Reject?  OK only if biorxiv.org is accepted by IEEE!! biorxiv.org pre-posted, other diverse authors <=5% similarity combined

********************
N-20318.pdf 	57.0%, 
<.IJCNN paper authors' names.>
<.IJCNN paper title.>

Overall view : arXiv paper by same authors
strong similarity (>=50%) : [References, ]
moderate (30% < similarity <50%) : [Research Status, ] 
low (<30%) :  [Abstract, Intro, III Identification model, Experimental, Conclusions]
My eyeball estimate of overall similarity (not including References) :  60%? (no need to do carefully)

49%	1 souce 47%, >20 all <= 3%

47%	http://export.arxiv.org/pdf/1810.12069
<.Other paper authors' names.>
<.Other paper title.>
>> same authors, same paper

>>>>> Howell :  N-20318.pdf 	57.0%, OK -  arXiv paper by same authors, other sources <=3% similarity

********************
N-20386.pdf 	52.0%, 
<.IJCNN paper authors' names.>
<.IJCNN paper title.>

Overall view : Previous author paper accounts for most of similarity, several paragraphs
strong similarity (>=50%) : [References, ]
moderate (30% < similarity <50%) : [oops ] 
low (<30%) :  [oops ]
My eyeball estimate of overall text similarity (not including References) : 
= sum(%similar*pagelength) / sum(pagelength)  
= / [100 * sum EACHBOTH *, sum second]  (0.75  0.25  0.6  0.2  0.2)  (0.66  1  1  0.4  0.25) 
= 44.5619

34%	1 source 34%, 45 sources 1% < sim <= 4% 

34%	https://ieeexplore.ieee.org/document/7727599
<.Other paper authors' names.>
<.Other paper title.>
>> primary author same, others different

3%		2 (same) sources 1% < sim <= 3%, 4 <= 1%

3%		https://academic.oup.com/jamia/article/25/3/321/4636780
<.Other paper authors' names.>
<.Other paper title.>
>> different authors, experimental description

>>>>> Howell :  N-20386.pdf 	52.0% (45%), Reject? Borderline case - my estimate of similarity is in acceptable range, also OK if high self-plagiarism in Introduction acceptable, >70% (?) of similarity is from authors' previous paper, paragraphs in Intro mostly, sentences elsewhere 

********************
N-20406.pdf 	52.0%, 
<.IJCNN paper authors' names.>
<.IJCNN paper title.>

Overall view : Previous author papers account for essentially all similarity, paragraphs & sentences
strong similarity (>=50%) : [III DN-2 framewk BC, III DN-2 Algorithm, Experiments, References, ]
moderate (30% < similarity <50%) : [Research Status, Conclusions, ] 
low (<30%) :  [Abstract, Intro, II Liberate symbols, III DN-2 A framewk , ]
My eyeball estimate of overall text similarity (not including References) : 
= sum(%similar*pagelength) / sum(pagelength) :  
= / [100 * sum EACHBOTH *, sum second]  (0.2  0.1  0.1  0.50  0.85  0.6  0.9  0.7) (0.6  0.95  0.75  0.85  1  1  0.7  0.6) 
= 49.845 

first group : 
35%	1 @ 35%, 1 @ 12%, rest <=4% mostly same authors 

35%	https://ieeexplore.ieee.org/document/8611300
<.Other paper authors' names.>
<.Other paper title.>
>> same authors 

12%	https://ieeexplore.ieee.org/document/8489455
<.Other paper authors' names.>
<.Other paper title.>
>> same authors

rest of groups :  <= 4%, same authors

>>>>> Howell :  N-20406.pdf 	52.0% (50%), Reject unless reword or cite own previous work,  Previous author papers account for essentially all similarity, paragraphs & sentences, other authors <=3% similarity

********************
N-20478.pdf 	51.0%, 
<.IJCNN paper authors' names.>
<.IJCNN paper title.>

Overall view : Previous author paper accounts for most of (very low) similarity, phrases mostly, rare sentences
strong similarity (>=50%) : [Abstract, II Related work, III Ball CDO, References, ]
moderate (30% < similarity <50%) : [Intro, ] 
low (<30%) :  [IV Spiking neuron, V Systm execution, Conclusions, ]
My eyeball estimate of overall text similarity (not including References) 
= sum(%similar*pagelength) / sum(pagelength) :  
= / [100 * sum EACHBOTH *, sum second]  (0.6  0.55  0.2  0  0.2) (0.6  1  0.8  0.4  0.15) 
= 37.2881

3 groups of sources @ [33%, 6%, 4%]

33%	>50 sources, 25 >= 5% similarity, 

33%	https://ieeexplore.ieee.org/document/8556772
<.Other paper authors' names.>
<.Other paper title.>
>> different authors

24%	https://ieeexplore.ieee.org/document/8268748
<.Other paper authors' names.>
<.Other paper title.>
>> different authors, same as 33% above

21%	https://ieeexplore.ieee.org/document/7946637
<.Other paper authors' names.>
<.Other paper title.>
>> different authors, similar to 33% above

many of remainder are from authors of 33% above

most other groups of sources refer to same authors of 33% above

>>>>> Howell :  N-20478.pdf 	51.0% (37%), Reject? Borderline case - my estimate of similarity is in acceptable range, most similarities are from other authors

********************
19Jan2019 11:28 - break as all >=50% have been done
come back for <50% later if I have the time




























# enddoc