Field Programmable Gate Array Applications—A Scientometric Review

Round 1

Reviewer 1 Report

Summary and Comments:

The authors investigate the more than 70000 publications on Field Programmable Gate Array (FPGA) found in the Web of Science and Scopus from 1992 to 2018 and merged into a uniform data set with duplicates removed, thereby covering this in the field of computational engineering highly interesting and very diversely applied technology from its start.

The authors identify about 150 top FPGA applications, which they group into 11 main categories. For each of these they review the development of the associated applications following a unified procedure, focusing primarily on detecting the year, when new topics have arisen, and if they have become trending topics in the most recent years.

To that end they make extensive use of their own recently published Python-based scientometric tool named ScientoPy, which is primarily concerned with temporal scientometric analysis. In particular they use two features of the tool: a two-fold evolution graph and the so-called trending bar graph:
Evolution of topics is displayed in two distinct graphs: the first one plotting the overall growth via the accumulative number of documents since the first occurrence of the topic up to 2018 and the second one in a parametric graph restricted to the last three years (2016-2018) by plotting an absolute indicator, the average document number per year (ADY), over a relative indicator, the proportion of the number of recently published papers over against the whole publication count over the life time of the
topic (PDLY). In some categories with more than 10 applications the authors only use a reduced presentation of the time evolution in form of bipartite bars for each application, indicating the absolute number of publications and the PDLY.

In general the present study seems to be an exhaustive search of the subject, with a reasonable structuring into 11 main categories and a suitable application of a new scientometric tool in order to display indicators that are interesting for those watching the upcoming technology subjects. For this purpose the extensive list of 1646 references could and should also be very helpful. The first about 60 references are listed for a more introductory purpose, but most of the remaining ones are connected
with the top applications, discussed in the different chapters with expert knowledge of computer science and engineering.

But there a weaknesses in the textual formulation of the paper, which should be edited again, preferably with the help of a native English speaker. I can only exemplify main recurring errors and suggest suitable corrections, indicating page number P and line number(s) L and the needed corrections by using “=>”. In some instances the meaning of the written text is unclear. I indicate these, too.

Corrections required/recommended:
P1, L1 (and others) “propose” => “purpose”.
P1, L32 Which are the “previous papers”?
P2. L50: “As since ...” => “Since ...”
P2, L51 and others: in the term “duplication removal filter” the word “filter” could and should be
dropped.
P3, L52 and others: “by ScientoPy preprocess” is one example of many, where the authors omit the
grammatical definite or indefinite article
P3, L60 and others: one example of grammatically incomplete sentences: “after the omitted papers”
(add: had been) “removed”.
P3, L60,66,67: examples of missing or incorrect genitive constructions: “the duplicated documents
number” => “the duplicated documents’ number”
P3, L63,66 and others: examples of incorrect use of “from”, should be “by”: “removal by the
preprocessing script” => “removal by the preprocessing script” etc.
P3, L65-67: misspellings, genitive, use of “from”: ScientoPy, DOI,
=>
“The duplication removal filter used by ScientoPy is based on the DOI match or if the DOI is not
present on the documents’ title and documents’ first author last name match.”
P4, Table 1: first column names not consistent with caption and/or explanations on page 3: it should
read “Total duplicated papers found” and “Total loaded papers”, “Loaded papers from Wos/Scopus”.
Caption line 2 : “Total duplicated papers found” must read “Total papers after omitted papers
removed”.
The term “total” should in any case be replaced by “total number of”.
Caption line 4: Where in the table are the “Duplicated documents with different cited by” listed?
P4, L68: where does the number 77699 come from. “Total papers after rem.dupl” are 77384.
genitive: “papers bibliometric information” => “papers’ bibliometric information”
P4, L70: What is the criterion for the “most relevant author keywords”?
P4, L73: “we analyze in deep” => “we analyze in depth”
P4, L83: “average documents” => “average number of documents”
P4, L93: Why do you use the term “S curve evolution”, when a sigmoid shape of the evolution of
topics is not predictable and even not found in the results?
P4, L104: The meaning of “cite of the papers ...” is unclear!
P4, L105: Does the sentence mean the following? “The papers most relevant for each application or
implementation are the most cited and newest ones for the topic.”
P4, L112: “PLY” => “PDLY”
P4, L113: “commit the requirements” => “comply with the requirements”
P5, L122: “in that way” => “in that”
P6, L165: one of some occurrences: “stared” => “started”
P7, L188: first of many examples for incorrect plural: “others” => “other”
P8, L221: “than we can be built” => “that can be built”
P11, L317: “arquitecture” => “architecture”
P12, L328: “traffic calcification” => “traffic classification”
P13, L359: “cathegory” => “category”
P14, L384: “headphones active noise cancellation” => “active noise cancellation in headphones”
P15, L419: “moving objects” => “moving objects detection”
P15, L424: “storage” =>“stored”
P15, L428: “storage” => “store”
P15, L426: according to my calculation : “22.39GB/h” => “22.39GB/min”
P16, L454f. Meaning of sentence “FPGA ...” unclear
P18, L508: “can change in or to an state that represents a ...” => “can change into a state that
represents a ...”
P22, L630: “enveloped” => “developed”
P22, L631: “growth” => “grown”

Author Response

Response to reviewer 1:

We thank you for these very positive comments, which acknowledge the importance and relevance of our manuscript.

Also, thank you for your useful and detailed review of our paper. We have answered each of your points below, and in the reviewed paper version.

English language and style: Moderate English changes required

- According to the reviewer's suggestion, English language and style were deeply reviewed in the full document by one of the English local authors.

Corrections required/recommended:

P1, L1 (and others) “propose” => “purpose”.

- According to the reviewer's suggestion, this and similar fixes were made.

P1, L32 Which are the “previous papers”?
- According to the reviewer's suggestion, this was clarified in the text.

P2. L50: “As since ...” => “Since ...”

- According to the reviewer's suggestion, this and similar fixes were made.

P2, L51 and others: in the term “duplication removal filter” the word “filter” could and should be
dropped.

- According to the reviewer's suggestion, this fix was made.

P3, L52 and others: “by ScientoPy preprocess” is one example of many, where the authors omit the
grammatical definite or indefinite article

- According to the reviewer's suggestion, this and similar fixes were made.

P3, L60 and others: one example of grammatically incomplete sentences: “after the omitted papers”
(add: had been) “removed”.

- According to the reviewer's suggestion, this and similar fixes were made.

P3, L60,66,67: examples of missing or incorrect genitive constructions: “the duplicated documents
number” => “the duplicated documents’ number”

- According to the reviewer's suggestion, this and similar fixes were made.

P3, L63,66 and others: examples of incorrect use of “from”, should be “by”: “removal by the
preprocessing script” => “removal by the preprocessing script” etc.

- According to the reviewer's suggestion, this and similar fixes were made.

P3, L65-67: misspellings, genitive, use of “from”: ScientoPy, DOI,
=>
“The duplication removal filter used by ScientoPy is based on the DOI match or if the DOI is not
present on the documents’ title and documents’ first author last name match.”

- According to the reviewer's suggestion, this and similar fixes were made.

P4, Table 1: first column names not consistent with caption and/or explanations on page 3: it should
read “Total duplicated papers found” and “Total loaded papers”, “Loaded papers from Wos/Scopus”.
Caption line 2 : “Total duplicated papers found” must read “Total papers after omitted papers
removed”.
The term “total” should in any case be replaced by “total number of”.
- According to the reviewer's suggestion, the names in table description, and paragraph were fixed

Caption line 4: Where in the table are the “Duplicated documents with different cited by” listed?
- This item were removed, because it is not relevant for this table.

P4, L68: where does the number 77699 come from. “Total papers after rem.dupl” are 77384.
this value in line P4, L68 was updated.

- The value in the paragraph was updated to the lasted value that is reported in the table

genitive: “papers bibliometric information” => “papers’ bibliometric information”

- According to the reviewer's suggestion, this and similar fixes were made.

P4, L70: What is the criterion for the “most relevant author keywords”?
- According to the reviewer's suggestion, we explain here that these are the author keywords related to FPGA's applications

P4, L73: “we analyze in deep” => “we analyze in depth”

- According to the reviewer's suggestion, this and similar fixes were made.

P4, L83: “average documents” => “average number of documents”

- According to the reviewer's suggestion, this and similar fixes were made.

P4, L93: Why do you use the term “S curve evolution”, when a sigmoid shape of the evolution of
topics is not predictable and even not found in the results?

- We removed the term S curve, because we agree that we can not predict the sigmoid shape of the topics.

P4, L104: The meaning of “cite of the papers ...” is unclear!

- According to the reviewer's suggestion, we clarified it by fixing to “the citation of the papers”

P4, L105: Does the sentence mean the following? “The papers most relevant for each application or
implementation are the most cited and newest ones for the topic.”

- It doesn’t mean that. It means that one of the criteria for citing a paper in this review is that this paper is one of the most cited in each application category. This was clarified in the reviewed version.

P4, L112: “PLY” => “PDLY”

- According to the suggestion, this fix was made.

P4, L113: “commit the requirements” => “comply with the requirements”

- According to the reviewer's suggestion, this fix was made.

P5, L122: “in that way” => “in that”

- According to the reviewer's suggestion, this fix was made.

P6, L165: one of some occurrences: “stared” => “started”

- According to the reviewer's suggestion, this and similar fixes were made.

P7, L188: first of many examples for incorrect plural: “others” => “other”
- According to the reviewer's suggestion, this and similar fixes were made.

P8, L221: “than we can be built” => “that can be built”

- According to the reviewer's suggestion, this fix was made.

P11, L317: “arquitecture” => “architecture”

- According to the reviewer's suggestion, this fix was made.

P12, L328: “traffic calcification” => “traffic classification”

- According to the reviewer's suggestion, this fix was made.

P13, L359: “cathegory” => “category”

- According to the reviewer's suggestion, this fix was made.

P14, L384: “headphones active noise cancellation” => “active noise cancellation in headphones”
- According to the reviewer's suggestion, this fix was made.

P15, L419: “moving objects” => “moving objects detection”
- According to the reviewer's suggestion, this fix was made.

P15, L424: “storage” =>“stored”
- According to the reviewer's suggestion, this fix was made.

P15, L428: “storage” => “store”

- According to the reviewer's suggestion, this fix was made.

P15, L426: according to my calculation : “22.39GB/h” => “22.39GB/min”
- Thanks to the reviewer's suggestion, we fixed this value to 1.2TB/h

P16, L454f. Meaning of sentence “FPGA ...” unclear

- According to the reviewer's suggestion, we clarified this sentence.

P18, L508: “can change in or to an state that represents a ...” => “can change into a state that
represents a ...”

- According to the reviewer's suggestion, this fix was made.

P22, L630: “enveloped” => “developed”

- According to the reviewer's suggestion, this fix was made.

P22, L631: “growth” => “grown”

- According to the reviewer's suggestion, this fix was made.

Reviewer 2 Report

The authors present a bibliometric study on publications dealing with FPGA. They searched the publications in Web of Science and Scopus and basically analyzed the annual number of publications. The results are presented for various topics in FPGA research which have been identified based on most frequent keywords in the FPGA publications. The study has been submitted to be published in a special issue on bibliometrics (and altmetrics). In my opinion, the study does not agree with the purpose intended with the special issue formulated by the guest editor (see https://0-www-mdpi-com.brum.beds.ac.uk/journal/computation/special_issues/bibliometrics) and should be rejected. In bibliometric (altmetric) studies, the application of advanced bibliometric methods is in the focus which is not the case with the current manuscript. It is not sufficient to present only annual publication numbers (or similar numbers), but to analyze citation impact, collaborations between authors, historical roots, public discussions, etc. using tools such as VOSViewer, CitNetExplorer, CRExplorer, CiteSpace, etc. For example, the guest editor of the special issue on bibliometrics published a typical bibliometric study on a certain topic. The study can be found here: https://0-doi-org.brum.beds.ac.uk/10.1371/journal.pone.0160393.

Author Response

Response to reviewer 2:

We thank you for your review of our paper. We have answered your main point below:

In my opinion, the study does not agree with the purpose intended with the special issue formulated by the guest editor (see https://0-www-mdpi-com.brum.beds.ac.uk/journal/computation/special_issues/bibliometrics) and should be rejected. In bibliometric (altmetric) studies, the application of advanced bibliometric methods is in the focus which is not the case with the current manuscript. It is not sufficient to present only annual publication numbers (or similar numbers), but to analyze citation impact, collaborations between authors, historical roots, public discussions, etc. using tools such as VOSViewer, CitNetExplorer, CRExplorer, CiteSpace, etc. For example, the guest editor of the special issue on bibliometrics published a typical bibliometric study on a certain topic. The study can be found here: https://0-doi-org.brum.beds.ac.uk/10.1371/journal.pone.0160393.

Response: We agree with the reviewer that bibliometric studies must include the analyze of citation impact, a collaboration between authors, historical roots, and others. Nevertheless, this paper is not a bibliometric study but it is a scientometric study, which is a kind of study valid for this special issue due it is included in the keywords' scope (see https://0-www-mdpi-com.brum.beds.ac.uk/journal/computation/special_issues/bibliometrics). In the scientometric study, the core is not the analysis of the bibliographic information, but here we have a review supported by the quantitative scientometric methodologies, due to the huge quantity of applications related to this study.

Reviewer 3 Report

Dear authors

Thank you for submitting this work where you provide a global vision of the applications of FPGAs. It’s my believe that this is an extraordinary contribution for folks that come now to the field of FPGAs as it permits to analyse the evaluation over the years, and can serve as a starting point for students, or instructors that are looking at applications. This is robust work that contributes to the state of the art and with a very high-quality presentation standards, the images are superb, is well-organized and the language is good. ScientoPy sounds like a very powerful tool, would love to know more and perhaps even use it myself for a future review. Congratulations on the tool that I’m sure was developed with great effort. I have but few comments to improve this work, see next:

Different literature reviews use different approaches, could you compare with main methodologies reported and the one that you have followed?

More explanation about how you group in categories is needed? What was the criteria? What happens with similar keywords?

I would separate Materials and Methods in subsections, right now is a bit messy and this can help with readability. My suggestion is “Review methodology”, “Data collection”, “Metrics”, but anything that makes it more readable can work.

Can you add a final paragraph on the discussion addressing the global picture of the impact of this review? E.g. how it can be used by teachers, students or researchers, and the potential that this kind of work can have in other areas of research as well.

Author Response

Response to reviewer 3:

We thank you for these very positive comments, which acknowledge the importance and relevance of our manuscript.

Also, thank you for your useful review of our paper. We have answered each of your points below, and included them in the reviewed paper version.

English language and style: English language and style are fine/minor spell check required

According to the reviewer's suggestion, English language and style was deeply reviewed in the full document by one of the English local authors.

Different literature reviews use different approaches, could you compare with main methodologies reported and the one that you have followed?

- According to the reviewer's suggestion, this scientometric study was compared with the main methodologies at the beging of the Review methodology subsection.

More explanation about how you group in categories is needed? What was the criteria? What happens with similar keywords?

- According to the reviewer's suggestion, more explanation was added here.

I would separate Materials and Methods in subsections, right now is a bit messy and this can help with readability. My suggestion is “Review methodology”, “Data collection”, “Metrics”, but anything that makes it more readable can work.

- According to the reviewer's suggestion, the Materials and Methods was divided in subsections.

Can you add a final paragraph on the discussion addressing the global picture of the impact of this review? E.g. how it can be used by teachers, students or researchers, and the potential that this kind of work can have in other areas of research as well.

- According to the reviewer's suggestion, the global picture of the impact was added.

Round 2

Reviewer 1 Report

The paper has substantially benefited from the redaction by a native speaker, and the expansion of most of the technical abbreviations is very helpful for the non-expert reader.

But there are still some minor (mainly orthographic) errors left, resp. have been newly introduced:

The line numbers are now the ones in the revised paper.

L17 and L38 “propose” => “purpose”.

L77 ScientyPy => ScientoPy

L98 prase => phrase

L184 rage => range

L387 planing => planning

L457 Infinite => , Infinite

L479 Deoxyribonucleic => Desoxyribonucleic

L501 355 MB/s => 356 MB/s # (24/8 ) * 1920 * 1080 * 60 /1024/1024 → 355.957

L596 represent => represents

L811 FPGAs based => FPGA-based

----

An additional strong recommendation for improving the paper:

As I understand, the categorization of the 5000 top author keywords had been done by examining them one by one with the authors’ expert knowledge and assigning them to predefined categories related to FPGA. It would be very interesting from a bibliometric point of view - which is the focus of this special issue – if the application of the established visualization tool VOSviewer would yield similar results. VOSviewer is able to read the SCOPUS-style output of the ScientoPy preprocessing script and to display a co-occurrence map of the author keywords. Using a thesaurus file to group synonyms should lead to a clustering of keywords that can be compared to the one done by the authors.

(Hint for the developers of ScientoPy: Unfortunately the output format of ScientoPy is not exactly like the one from SCOPUS, which would be strongly desirable, for now the VOSviewer is not able to draw valid co-authorship maps on the basis of author, organization or country.
The authors field must be enclosed in double quotes, the authors must be separated by commas, and the author names must be written as surname plus initials with dot - like , e.g., “Tolkien J.R.R., Lewis C.S., Kennedy J.F.”. For the other units even more would have to be adapted.)

Author Response

We thank you for these very positive comments, and the suggestion to improve the quality of our paper. Also, thank you for your useful and detailed review of our paper. We have answered each of your points below, and in the reviewed paper version.

L17 and L38 “propose” => “purpose”.

- According to the reviewer's suggestion, this fix was made.

L77 ScientyPy => ScientoPy

- According to the reviewer's suggestion, this fix was made.

L98 prase => phrase

- According to the reviewer's suggestion, this fix was made.

L184 rage => range

- According to the reviewer's suggestion, this fix was made.

L387 planing => planning

- According to the reviewer's suggestion, this fix was made.

L457 Infinite => , Infinite

- According to the reviewer's suggestion, this fix was made.

L479 Deoxyribonucleic => Desoxyribonucleic

- According to the reviewer's suggestion, this fix was made.

L501 355 MB/s => 356 MB/s # (24/8 ) * 1920 * 1080 * 60 /1024/1024 → 355.957

- According to the reviewer's suggestion, this fix was made.

L596 represent => represents

- According to the reviewer's suggestion, this fix was made.

L811 FPGAs based => FPGA-based

- According to the reviewer's suggestion, this fix was made.

As I understand, the categorization of the 5000 top author keywords had been done by examining them one by one with the authors’ expert knowledge and assigning them to predefined categories related to FPGA. It would be very interesting from a bibliometric point of view - which is the focus of this special issue – if the application of the established visualization tool VOSviewer would yield similar results. VOSviewer is able to read the SCOPUS-style output of the ScientoPy preprocessing script and to display a co-occurrence map of the author keywords. Using a thesaurus file to group synonyms should lead to a clustering of keywords that can be compared to the one done by the authors. (Hint for the developers of ScientoPy: Unfortunately the output format of ScientoPy is not exactly like the one from SCOPUS, which would be strongly desirable, for now the VOSviewer is not able to draw valid co-authorship maps on the basis of author, organization or country.

The authors field must be enclosed in double quotes, the authors must be separated by commas, and the author names must be written as surname plus initials with dot - like , e.g., “Tolkien J.R.R., Lewis C.S., Kennedy J.F.”. For the other units even more would have to be adapted.)

- According to the reviewer’s suggestion, we improved ScientoPy to export the pre-processed authors’ names data according to Scopus format. In this way, we were able to use VOSviewer to generate a network map of the author keywords using the merge data from Scopus and WoS. In this way, a new section called “Applications co-relations” was added to describe the main co-occurrence map of the main FPGAs’ applications.

Reviewer 3 Report

I think the authors have further improved the manuscript based on the previous comments and it's ready to be accepted. Again, this is robust work that contributes to the state of the art and with a very high-quality presentation standards, the images are superb, is well-organized and the language is good.

Wishing the best to the authors in their future endeavours where I hope we can cross paths again.

Author Response

We thank you for these very positive comments, which acknowledge the importance and relevance of our manuscript. We are glad to know that you think that this paper is ready to be accepted.