Check-lists for publications

What is this document?

This document contains important guidelines to be followed while submitting papers or making presentations.

First version: May 2018. Last update: February 2022.

Paper writing

Questions especially towards the principal authors who has implemented the model and/or executed the experiments:

  1. Do you use exactly the same evaluation procedure and the performance metrics in your comparisons, especially against the state-of-the-art?
    • A good practice here is to reproduce the experimental results of a state-of-the-art system, and use the same code-base for running your experiments. (If you obtain signficantly lower or higher results, the chances are that you are doing something differently or wrong in your evaluation.)
  2. Is the information contained within the paper sufficient for re-producing the experimental results?
    • If not: is the missing information really so low-level that we can leave it out? Or should it be added to the manuscript?
  3. Do you think a part of the paper is not clear enough?
  4. Do you think a part of the paper is misleading or inaccurate?
  5. Do you see any inconsistencies between what is written and/or implied in the description of the method versus the source code?
  6. Do you see any inconsistencies between what is written and/or implied in the description of the experiments versus the source code?
  7. Do you feel confident and comfortable in releasing your source code upon publication? (If not, why??)
  8. Do you see any inconsistency in the experimental results?
  9. Do you think any one of the results is unexpected or unusual?

Whenever you see a problem, either fix it (if the solution is straight-forward) or inform all co-authors clearly and explicitly about the problem, so that, a proper action can be taken. Never overlook any detail!

Be aware that as the principal programmer and/or (co)author, it is primarily your responsibility to ensure the correctness and completeness of the paper contents.

Paper submissions (conference or journal)

  1. Prepare a full draft at least 1 week before the deadline. Papers failing to meet this goal tend to be buggy and/or incomplete.
  2. Avoid submitting any preliminary work, ie. avoid submissions with incomplete text, incomplete experimental analysis or those with results that are not fully confirmed (see the questions above).
  3. Ask all co-authors if there are research grants and compute infrastructures that should be acknowledged. This step is especially crucial for the final version of the submissions. Hide this information if the submission is double-blind.
  4. Spell-check using aspell -c filename.
  5. Grammar-check using pandoc --to=plain --wrap=none paper.tex plus a decent word processor (Google Docs, MS Word, etc).
  6. Obey the page limit rules.
  7. Preferably use bitbucket or github private repository for collaboration.
  8. Ask all co-authors their preferred author names, eg. I use "Ramazan Gokberk Cinbis".
  9. Make sure all information is consistent between the paper and the submission website, including the paper ID, title, abstract, and, the author names (if not double-blind).

arXiv submissions

  1. Follow the "Paper submissions" section first.
  2. Reminder: note that even when grant and compute infrastructure acknowledgments are absent from the original manuscript under-review, they must still be included in the arXiv version.
  3. Use the "I grant arXiv.org a perpetual, non-exclusive license to distribute this article." option for the license.
  4. Make sure the conference / journal allows submitting the paper to arXiv.
  5. Clean up latex folder (delete unused files). Never submit the publisher-edited pdf. Always use LaTeX source for submission.
  6. Papers under review: Use preprint (if exists) or final option when compiling the paper. Remove elements that may violate double-blind rules, such as paper ID.
  7. Accepted papers: Use final option when compiling the paper.
  8. Journal papers: Carefully read and follow the publisher policy: IEEE, Springer, or, Elsevier.
  9. Once the submission is complete, forward all arXiv emails and the arXiv-generated pdf to your co-authors.

Thesis submissions

Ask Gokberk for the detailed guideline doc.

Conference and thesis presentations (oral or poster)

  1. Read the excellent suggestions by Ranjit Jhala.
  2. Obey the time limit rules.
  3. Do several self-practice talks, and arrange at least one practice talk with your team-mates.
  4. Ask all co-authors if there are research grants and compute infrastructures to be acknowledged.
  5. Do not use TUBITAK logo (see rules).
  6. Give references directly within each slide, not on a separate slide.
  7. Use slide numbers.
License