How to write a research paper. A step by step guide:
Step 1: Have something to write about
Students sometimes forget this step. In the rat race to publish papers, especially in computing, graduate students start their research with the paper draft in mind already.
Don’t do that.
First, do the literature review. Print papers. Keep notes. Form a bibliography and generate a survey.
You can write that survey out if you want and make that its own paper, but that sometimes comes later.
Find a gap in the research, fill that gap with a study or innovation.
Collect data, perform statistics, form conclusions.
After all this is done, then (and only then) should you…
Step 2: Architect the Paper
Create an argument scaffolding.
Write from outside in, do not write from top to bottom.
What are the sections of your paper? What goes inside each section?
Then write the first sentence of the main paragraphs for each section.
It is often best to use a figure rather than prose. So architect your figures after you architect your prose. Oftentimes this is a rough hand-drawn figure that I copy and paste into the document as a placeholder
I often pre-design results figures in the same way.
Step 3: Write some supporting sentences
Each topic sentence needs to be supported with additional prose. This is time consuming.
Cite relevant papers and flesh-out the story
Step 4: Create Illustrations
Figures and tables are critical to the telling of the story. Figures must be polished to the highest possible publication standards.
Know and understand the differences between vector graphics and raster images. Use vector-based illustrations in all instances except live-action photos.
Don’t assume that a PDF is a vector-graphic.
Never, ever use raster images for data plots.
Step 5: Put data, complete statistical analysis, and appendix online
Modern research papers almost always point to datasets and statistical analysis somewhere online. Online appendices and supplementary materials are also posted online for additional references.
Extra credit if a reader can download your python notebook and data. Click run. And recreate all your statistics and plots. — some say that this ought to be required for empirical papers. I tend to agree.
Step 6: Take a nap
At this point you’ll have done a ton of work. And you’ll start experiencing semantic satiation. This is like a paper-writing version of a mirage in the desert.
Take a break for a few days, read a book, binge a TV series, have dinners with friends.
Step 7: Polish the paper
With fresh eyes, go over the paper carefully. Add in additional supporting narratives. Cut parts that don’t belong.
Make the paper fit to size constraints.
Make sure there are no dangling words or weird latex artifacts.
A research paper should look professional. A hastily crafted research paper will not be published in any respectable venue.
Computing journals and venues promote quick writing and publishing, but this is not the norm. Other scientists like sociology, law, and political science write working papers that spend years in development before they’re published.
Step 8: Submit it and Forget it
Once you submit you’ll be anxious to get your reviews back. Conference papers often have a review date, where the organizers will try to get the reviews and decisions back to the authors.
Ignore that date. Don’t stress over it. Move on to the next thing and let the email surprise you.
Decisions are rarely on time. Move on to the next thing. If you agonize over the decision, then you’ll waste a lot of time.
Step 9: Read the reviewers’ comments
So you’ve receiving reviewer comments. They hate your paper.
If the paper was rejected, you’ve got work to do. Read the reviewer comments, but don’t take it personally. This is hard. Reviewer comments are a tough pill to swallow. Carefully read the comments.
(I have a hard time doing this, personally)
Step 10: Makes changes
If your paper was rejected, then you’ll need to make changes and try again at another venue or revise and resubmit to the same venue at the next opportunity.
If your paper was accepted, then you’ll still need to make changes. Reviewer feedback can be helpful even if the paper is slated for acceptance.
Take the reviewers comments seriously, but remember that you are the author of the paper, not the reviewers. It is completely find to disagree with a reviewers remark and refuse to make some suggested change.
Step 11: Submit the final camera ready copy
The instructions for how to do this vary widely. Oftentimes a copy editor will confirm a few things and make changes. They will confirm all this with you.
There will sometimes be license and copyright issues.
There will sometimes be open access fees or conference registration fees that are required.
Step 12: After its published
GOTO Step 1