How to Write Research Papers

I'm a Ph. D student working on robotics and deep learning. Like most researchers, I also need to publish research papers. Preparing research papers takes much effort and time. Sometimes it even gives pains. However, I realized that there are a bunch of materials on the web I can benefit from and started summarizing those materials. This document might contain disputable statements since it is for my personal retrieval. I welcome comments and collaboration. Please send me e-mails or pull requests.

Key Things to Keep in Mind

Write Now!

• "Writing first" forces you to be clear & focused
• It helps you develop the idea
• It makes what you don't understand concrete

Top Down

• Identify your key idea and contributions
  • Just one clear and sharp idea
  • If you have lots of ideas, write lots of papers
• Tell a story
  • Not what you did, but rather new idea or insights
  • Basic flow
    • Here is a problem
    • It's interesting and unsolved
    • Here is my idea
    • My idea works (details or data)
    • How my idea differs from others
• A lead sentence to set context for a paragraph

Reader Centered

• Get to the message ASAP
  • Convey the intuition primarily; details are secondary
  • State your contributions clearly
• Visualize (figures) & summarize (tables)
• Don't let readers work hard
  • Be consise, accurate and clear; present enough context & information
  • Not too many equations or graphs or paragraphs
• Check what readers might (want/not want) and (know/not know)

Take Time

• Reflect, write, review, and refine
• Take time to write less

Where to Focus

• Title (1000 readers)
• Abstract (4 sentences, 100 readers)
• Introduction (1 Page, 100 readers)
• The problem (1 page, 10 readers)
• My idea (2 pages, 10 readers)
• The details (5 pages, 3 readers)
• Related work (1-2 pages, 10 readers)
• Conclusion and future works (0.5 pages)

How to Write Actually

Title

• Catchy buy avoid being cute
• Should represent the key idea and contributions

Abstract

• Describe the work not the paper
• Present concrete results rather than abstractive desription

1. General problem space/motivation
2. Approach taken
3. Key findings/results
4. Why the findings/results matter

Introduction

• Motivation for the work
• Prepare readers for the structure of the paper
• Start broad and progrssively narrow down to the issue (problem)

1. (Context) State of the world: the problem area & why it is important
2. (Need) Narrow down: gap between actual and desired situations (the big BUT)
3. (Task and Object) To achieve (object), we study/develop/investigate/design/evaluate/implement/assess ...
4. (Outcome) The key results are ...
5. (Contribution) In summary, the contributions of this paper are ...
6. (Structure) The rest of this paper is structured/organized as follows.

=> Need a section for "related work"

or ...

1. (Context) State of the world: the problem area & why it is important
2. (Problem) Narrow down: the desired goal which is not achieved
3. (Previous Works) Approaches taken in previous works and limitations (gap between actual and desired situations)
4. (Task and Object) To overcome the above-mentioned limitations, we ...
5. (Outcome) The key results are ...
6. (Contribution) In summary, the contributions of this paper are ...
7. (Structure) The rest of this paper is structured/organized as follows.

Materials and Methods

• Sufficient detail for reproducibility

Results and Discussion

• Interprete the results

Conclusion

• Present the outcome at a higer level of abstraction

Userfule Tools

• Thesaurus: Thesaurus, Onelook
• Grammar & Style: Grammarly, HemingwayApp

References

• Professor Kim's lecture note
• How to write a great research paper
• Top-10 tips for writing a paper
• Writing scientific papers
• The elements of style