Computing and Electrical and Electronic Engineering: Republication of Conference Papers

Abstract

The republication of conference proceedings as journal articles seems to be particularly prevalent in the areas of Computing and Electrical and Electronic Engineering. Our own use of CrossCheck turned up a disproportionate amount of duplication of this type in these particular disciplines; we also noted in our previous survey (see Chap. 2) that the attitudes to this question of authors in this area were markedly different from those in other fields. We therefore felt that it would be useful to carry out a survey of the attitudes of the editors of relevant journals to this issue; we surveyed over 300 journals in the field, and the results were published in an article entitled ‘Republication of conference papers in journals?’, which is reproduced in full (with the publisher’s permission) below [1].

Appendices: Free Text Responses to Question 3 and Question 4

Appendix 1: Free Text Responses to Question 3

No.	Q3-Pt1: Please outline the nature of the changes (n = 91)	Q3-Pt2: Why do you consider that these changes are necessary? (n = 87)
1	Should be 70 % changes to the paper published in conference proceedings. The changes should include detail out related work, further evaluation of method	No point of having the same article. The most important thing is the dissemination of knowledge
2	At least 30 % across most aspects	The journal paper should be an improvement of the conference one
3	Expanding out all omitted materials in the Proceedings due to space constraints	The journal paper should be complete
4	30 % new material	Conference proceedings are now available on line
5	The overlap shouldn’t be more than 50 %	The journal version of a conference paper should include much more details and insights
6	At least 30 % new materials added	Why publish the same paper in the different forms?
7	Substantial additions, more thorough coverage, results, comparisons, etc.	So that the reader will benefit from reading the paper (having read the conference proceedings)
8	The paper must contain contents largely expanded to cover the advancement in the topical subject or field of study, typically with 70 % percent more or different materials compared to the conference paper	Mandatory!
9	Significantly more experiments, explanation, and algorithmic details	There is no need to publish the same paper twice. The journal provides a longer format for more complete presentations; that is its purpose
10	Improve the quality of the papers, 30 % new	Conference papers are inevitably in complete and of lower quality than what are needed for a journal article
11	Add 30 % new content	Explain the work in more details, deepen the theory and report more experimental results
12	The journal submission must contain 50 % new materials
13	I believe that an Extension with more details, theory, examples, concepts, results, experiments and comparisons are required
14	At least 20 % original material	To respect the copyright
15	The journal paper needs to be a mature, well-founded contribution that has been fully validated either theoretically or experimentally. Usually, the conference paper only outlines the contribution, thus it needs to be substantially Extended with proofs and/or Complete experimentation	To obtain a Complete self-standing contribution for the future, which other works can build on
16	There has to be at 30 % new material in the journal version	There should be added value otherwise the authors are simply publishing the same paper twice
17	To have more Complete version and to improve quality	For journal should have more complete and better quality papers
18	Detailed proofs and full explanation should be given	A conference paper (with page limit) considered as an Extended abstract should be published in a Journal in a Complete form. My area is mathematical, and journal review is mandatory to be recorded as a correct result in the history
19	Substantial extension, new experiments, etc.	Otherwise it is a publication duplicate, this should never be accepted!
20	Expand and update discussion	These changes make the paper better; conference proceedings have unnatural constraints
21	1. At least 25 % new material 2. Revisions requested by reviewers	(1) is the policy of the Association for Computing Machinery
22	Significant changes/enhancements (min 30 % enhancements, often more)—both conceptually and in details/evaluations. Publications must add substantial value adds	We really do not Republish the papers but would consider papers that have substantial new material that builds upon a prior conference/workshop paper
23	Extended research results, validation, evaluation	Substantiate more early research results
24	Submissions can contain material published in one or more conference papers. Submissions should contain at least 20 % percent by length of material that is unpublished. This new material should make substantial new contributions. This would typically be additional theoretical or experimental results, or both	Strengthen the quality and significance of the previously published material
25	Major Extensions, and avoid self-plagiarism	A paper published in a conference is to be considered archival. Also, given widespread access to internet, it is easily available. Therefore, no reason to duplicate publications
26	We require significant Extension, such as new results and/or discussion, to provide substantial 30 % ‘value added’ to the journal version. We also require that the paper be revised, if necessary, to be sufficiently complete and thoroughly documented for Archival journal publication	The conference paper is already published and available. In addition, the standards for journal publication are higher
27	The title may be the same, but the journal submission must be much more detailed and complete than a conference presentation. We rely on the opinion of expert reviewers to determine if the submission is worthy of consideration	Permanent archival journals require rigorous attention to details to satisfy expert reviewers. Almost nothing is accepted without a major revision
28	The proceedings should be an extended abstract	Otherwise it would be basically as in the proceedings and then the answer to 1 would be NO
29	Extension in theory, algorithm, and/or empirical validation
30	Complete proofs of results, improvements after conference feedbacks	To have a self-contained presentation of results
31	Full proofs of Theorems, more complete examples, any additional material that could not be included in the conference paper due to strict page limits	Republication is allowed only because conferences impose page limits
32	Substantial changes with at least 30 % new material	Conference publications are already peer reviewed publications. To republish substantially the same work would not be honest, and would devalue the original publication
33	Must be significantly different with the conference paper	Conference papers are simple and quick publication, journals are more formal and for wider distribution and reference
34	More depth or additional material	Conference reviews are patchy, page limits are tight
35	conf proc are preliminary reports, to be refined for peer review; if we have to quantify … 40 % reuse is a threshold;	Goal is to preserve conferences as a place to discuss research in process; reviews are rarely as thorough as for a journal; expected that authors needs to revise
36	Substantial expansion of new content	So the journal version represents added value to the reader
37	Additional theoretical and empirical results, additional consideration of related work	Conference proceedings are Archival in nature in CS; the same paper should not be published twice, and the journal’s role is to publish expanded, more thoroughly-developed versions of the research
38	Longer paper, more results, feedback from conference	To make the idea more Complete
39	Improvement in contents and changes in writing	Yes
40	The paper must be a significant extension of a conference paper	We do not allow duplicate publication
41	60 % must be new	We cannot publish the same paper
42	Substantial changes and extension of the conference version, 30 % new	Conference papers in CS are proper publications—copyright resides with the publisher in many cases
43	At least 30 % new material	Re-publication of the same material would be the same as publishing the same paper in two journals: unfair and unethical. From a practical standpoint, it would also make the journal superfluous
44	The journal publication gives authors a chance to add details, proofs, and comments, to Expand and to polish their presentations, and to fix typos and errors, inevitable for deep works presented by deadline to the conferences, where no corrigenda are available and the reviewers have too little time; the authors should use this chance; besides I should have some support to counter possible objections from some excessively zealous reviewers who cannot see the need for journal publication of proceedings papers and blindly apply the copyright law, not caring about the best interests of the field	The reasons are explained above
45	We require sufficient changes that someone finding both the journal article and the conference publication would receive value from the journal article over the conference publication. In other words, there has to be sufficient additional, new material to justify publication in the journal	To justify the second publication
46	Substantial amount (e.g. 30 %) of new material/work	Not to republish things already published
47	Significant new and/or expanded material must be included	To avoid duplicative publication of essentially the same material
48	I’m answering this as if I were EIC of IEEE S&P Magazine, which I was until January 2011. This publication is a magazine, not a journal. Hence it aims to publish papers of general interest and not, in general, cutting edge research. It is subject to the IEEE’s rules on republication. Usually, articles that are republished in the magazine require shortening and editing to make them suit the magazine’s audience	The changes are needed to tailor the article’s content to the magazine’s audience
49	The article must deal the topic more in details, with new results	While journal articles have a wider diffusion, conferences (especially the main ones) a becoming more and more accessible via the web
50	Some technical details	Quality
51	Substantial Extension	Conference papers generally are short and do not have the breadth of a journal paper
52	Substantially new technical material, >60 %	Yes
53	30 % new material	Added value of the new publication
54	Give more explanations if needed	To overcome the page limit with conferences; and because often sufficient details are needed to make the work reproducible by readers
55	At least the paper should be 75 % different from the one published	Otherwise, this constitutes double submission
56	The paper is sent to referees for a more formal review by experts in the subject area	Typically, the conference proceeding version has been quickly reviewed by one editor, but did not go through the typical detailed review process
57	At least 30 % Substantive additional content	Otherwise no additional value is obtained from republishing the material
58	30 % extra material	Republishing only existing material gives no added value over the conference publication
59	Min. two new reviewers of the journal will check the improved conference paper again. If, for example, the paper was originally published at an ASME conference proceeding, then no copyright issues occur in case of re-publication in an ASME journal. I do not accept a conference paper that was published by another publisher	Usually, the review processes of conference proceedings papers are not as strict as the journal reviews, also the page limits and the format requirements are different. Actually, someone may submit a conference paper unchanged, but to have 2 papers with the same title and same content is strange—the authors would like to carry out some changes after having the experience at the conference
60	Add new material (at least 30 %, 1/3 of the paper must be new material)	Conference papers are typically shorter
61	It depends on the paper! More details of proofs, more experiment reports, development of the state of the art, etc.	Conference papers are too short to develop comprehensive presentation
62	30 % new material
63	At least 30 % new material, significant results of lasting value	The material has already been published—same reason we do not republish articles from other journals. Also, copyright in at least some cases
64	Higher standard for acceptation	Higher quality
65	Substantial changes at least 40 % new content	Because the paper should be substantially different
66	Addition of significant new content	Otherwise a separate publication is not justified
67	Extension and depth	To improve the quality
68	There should be at least 50 % new material in the journal submission	To avoid copyright issues and also to emphasize that the motivation for conference papers and journal papers are different and their styles should be different as well as the amount of detail provided
69	30 % new material, either in the form of new theoretical material, or new experimental results	The journal version should be more extensive than the conference version
70	Extended results, e.g. more experiments, plus Extended related work, plus a more extensive description of the presented framework, or tool, or approach	The conference proceedings is already a referable publication per se, it makes no sense having it identically republished elsewhere
71	Provide full proofs of the results (which usually have been omitted from the conference version due to space limitations)	Journal publications are considered the definite publication of a result and must contain full proofs
72	Fuller account of data and methods, new analyses, and original text (i.e. not cut-and-paste)	Yes
73	In order to ensure copyright for both the conference and Elsevier we insist on a major rework—usually making the paper longer and more technical	Legal requirements and ensuring quality papers
74	Paper should add more in depth information; paper should contain >50 % new material compared with the conference	We should avoid duplication (I mean copy and paste). Extended paper should add more value, more information, and more details. I do not accept cosmetic changes!
75	We will not publish full papers from conf. proceedings (usually ACM, or Springer LNCS) but will accept Substantially modified and extended submissions of Work-In-Progress, poster, short papers (<4 pages) published as part of such proceedings	The aim of ‘short’ submissions is different from that of journal articles but often contains up-to-the-minute work which can be put in context and expanded upon in a journal submission. Does not really count as a ‘citable’ publication as is often in an Appendix or 2nd volume of related conference publications
76	Typically require more experiments, proofs, details of implementation, additional to previous work	Conference publications tend to be short and therefore in complete. They also tend to be preliminary in nature, sometimes errors are present. The conference review process is hurried and light, concentrating on innovation rather than long-lasting significance or importance. Change for the sake of change is not necessary; thorough presentation of the material is what typically creates the need for changes to the conference paper in order to make it suitable for journal publication
77	New results must be included	To produce new results for researchers
78	I expect conference proceedings papers to be short versions of what would be more complete papers	I would not simply republish a conference paper, the journal paper would have to be more Substantive and contain a more Complete description of the work and additional results
79	More precise explanations, theoretical background, case studies, language improvements, etc.	Conference papers have obviously limited number of pages and quality of journal papers should be better than conference ones
80	Significant improvement or extension	Duplication adds no value
81	Extensions in contribution with clear explanation of the novelty of the ‘new’ paper, 50 %	We should not support re-publication of research; we should support publication of new and novel research contributions. There may also be copyright issues if re-publishing as well as problems with which version to cite. As an editor, I want other researchers to cite the paper in my journal, and I believe this is best achieved if a paper has an added value in relation to a previous publication
82	Should be at least 30 % new material	There is no point having the same paper twice
83	The submission should contain a substantial amount of new results	To make the paper strong enough for journal publication
84	Papers have to be considerably expanded	for solving copyright issues
85	At least 30 % new material (RESULTS), different name, original paper needs to be cited and differences explained	Re-publishing the same paper would be duplicating content that is already available in ACM or IEEE digital libraries
86	Extend the paper	Journal paper should be more consistent
87	30 % more material	The conference version is already peer reviewed and archived
88	A different perspective and angle would generally be required or more detail	There is no point having an identical copy of the paper everywhere; problem is papers as conf articles are not cited or read anywhere near level of journals so journals are mandatory. Conferences are clearly important for interactive dialog
89	Provide full proofs of claims	Because journal publications should contain all details necessary to understand claimed results
90	30 % new material	It’s common practice
91	More than 40 % new results and outcome (qualitatively)	My journal policy is set to not allow to have double archives documents, and also it is not appropriate to republish similar things

Appendix 2: Free-Text Responses to Question 4 (n = 27 Journals Editors)

No.	If NO to Q1, why not?
1	Because it has already been published. We only accept such papers if it can be demonstrated that more than 50 % of the material is new
2	A paper should not be published twice, in different media. Archival publication in a journal is a different topic than a conference paper. Of course, the journal article can speak about the same work presented in the conference, but the journal paper is of a very different nature and should never be a republication
3	Duplicate publication
4	We publish only original contributions
5	These would be part of special issues based on the best papers in a specific conference. We do not accept individual papers which have been published in general conferences
6	Conference papers are easily accessible on the web. No need to republish
7	Our policy is here: http://www.ida.liu.se/ext/aijd/authors.html—AIJ considers eligible for publication mature high quality work that is Complete and novel. The question of whether a paper is complete is ultimately determined by reviewers and editors on a case-by-case basis Generally, a paper should include all relevant proofs and/or experimental data, a thorough discussion of connections with the existing literature, and a convincing discussion of the motivations and implications of the presented work. A paper is novel if the results it describes were not previously published by other authors, and were not previously published by the same authors in any archival journal. In particular, a previous conference publication by the same authors does not disqualify a submission on the grounds of novelty. However, it is rarely the case that conference papers satisfy the completeness criterion without the addition of new material. Indeed, even prize winning papers from major conferences often undergo major revision following referee comments before being accepted to AIJ. Authors should ensure that they have appropriate copyright permissions before submitting any material that has been previously published
8*	Editor 1. It is no longer original. It would only be accepted if a significant amount of additional material were added, such as reviewing other work, and a full evaluation. SO it would be ok if the conf paper formed the basis of a journal paper, but it would never be published ‘as is’ if the conference already had published it
9*	Editor 2. They have already published (*No.8 and No.9 in same journal)
10	Since conference proceedings are already widely available on the internet, there is no added value to having the same content appear in a journal as well
11	We follow the IEEE policy on self-plagiarism
12	Often conference proceedings require transfer of copyright, and thus we cannot publish the same material
13	In my field, conference proceedings are often considered archival, so publishing a conference paper again in a journal would be duplicate publication. We do allow articles that consolidate and extend work published previously in conference proceedings, however
14	Published papers should be updated, following suggestions from the audience
15	We prefer original conceptual papers which are innovative
16	Because they are already available. We would typically set up special journal issues, peer-review the papers, and published the ones that were acceptable
17	It is not necessary
18	We only consider full length original research articles
19	We publish ‘original research’
20	Duplication
21*	Editor 1: It’s a silly question. Common practice is to publish a paper RELATED to a conference paper, but having sufficient changes to be considered a new paper. (Otherwise the conference publisher would own the copyright and we wouldn’t be able to republish any way!) Typically, papers need 40 % or more of new material. They also need to be more rigorous than conference papers in various other ways, e.g., we’d normally expect all formal claims to be proved and all empirical claims to have evidence provided for them
22*	Editor 2: Publication is warranted only once—republishing material serves no legitimate purpose, and only inflates a person’s list of publications with misleading information. (*No. 21 and No. 22 in same journal)
23	The journal is dedicated to publication of new results—not recycling of old results with minor updates
24	Too many authors then count the paper twice. It’s a con. There is the occasional exception where the paper gets radically revised and becomes in effect a new paper, but most conference papers are not like this
25	It is necessary for authors to submit new papers that are derivative of their prior conference papers. Publishers (ACM vs. IEEE vs. Wiley) have policies and concerns, so we try to not cross them
26	The only reason for a journal publication is to provide a stamp of being reviewed. Conference papers are usually not thoroughly reviewed, in particular in terms of correctness
27	My journal policy is set to not allow to have double archives documents, and also it is not appropriate to republish similar things