Keynotes for Paper Writing

Table of Contents

General tips for writing


It should be clear how each sentence and paragraph relates to the adjacent ones.


  • Begin sentences with old information. Create link to earlier text
  • End sentence with new information. Create link to the text that follows, also places new info in position of emphasis


It should be clear how each sentence and paragraph relates to the big picture.

Rule2: One paragraph, one point.

  • A paragraph should have one main point, express in a single point sentence.
  • Typically the point sentence should appear at or near the beginning of the paragraph.

Name your baby

Rule3: Give unique names to things and use them consistently

Just in time

Rule4: Give information precisely when it is needed, not before.

Structure of a research paper

The basic idea: TOP-DOWN.

Explain your work at multiple levels of abstraction starting at a high level and getting progressively more detailed. A typical structure is:

  • Abstract(1-2 paragraphs, 1000 readers)
  • Intro(1-2 pages, 100 readers)
  • Key ideas(2-3 pages, 50 readers)
  • Technical meat(4-6 pages, 5 readers)
  • Related work(1-2 pages, 100 readers)


The CGI model for an abstact/intro.

Context. Set the stage, motivate the general topic.

Gap. Explain your specific problem and why existing work does not adequately solve it.

Innovation. State what you’ve done that is new and explain how it helps fill the gap.


There are two ways in writing the introduction.

  • Like an expanded version of abstract
  • Eliminate Context. For example:
    1. Start with a contrete example.
    2. If this works, it can be effective, but I find it often doesn’t work.

In the second way, it assumes reader already knows the context.

Another outline of introduction section:

  • Elevator story – high-level statement of the problem
  • The problem in more technical terms
  • Conventional wisdom: sketch of previous techniques and their shortcomings
  • Describe the solution to the problem as a black box, so that it is clear what your solution offers over previous work
  • Technical challenges to obtaining a solution (e.g., 1 paragraph for each)
  • State how you solve the challenges (at most a few paragraphs)
  • Experimental highlights
  • Contributions
  • Organization of the paper
  1. Use an example to introduce the problem
  2. Use a list of contribution. 3~5 bullet points. 2 is too little, >5 will make more things to criticize.
  3. Describe the contribution refutably. (e.g., We give the syntax and semantics of a language that supports concurrent processes (Section 3). Its innovative features are…) wrong: We describe the WizWoz system. It is really cool.
  4. No related work yet! Readers do not know the context, the background and the terminology, so do not talk about the tradeoffs, it is absolutely incomprehensible. Except:
    • Sometimes it is better to discuss the 1-3 most important pieces of related work
    • The best way to introduce your idea might be to contrast it with some well-known concept
    • Sometimes you need to make sure that the reader understands some essential background on which your work depends
    • Want to ensure that the reviewer does not dismiss vour idea as “It’s just the same as XXX.”

Key ideas

  1. Use concrete illustrative examples and high-level intuition.

  2. Do not have to show the general solution.(that’s what the technical section is for)

Why have a “key ideas” section at all?

  • Force you to have a “takeaway”
  • Many readers only care about the takeaway, not the technical details
  • For those who want the technical details, the key ideas are still useful as “scaffolding”.

Limitation/Threats to Validity

Construct validity

  • Are the parameters studied in the experiments relevant to the research questions posed?

Internal validity

  • Is there selection bias in the examples chosen for study?

External validity

  • Do the conclusions hold in other scenarios

Ideally: explain how you addressed or controlled for these issues

It goes at the end of the paper. You can only properly compare to related work once you’ve explained your own.

Give real comparisons, not a “laundry list”! Explain in detail how your work fills the Gap in a way that related work doesn’t.


  1. How to write papers so people can read them

  2. Tips on Writing a Research Paper

Aoyang Fang
Aoyang Fang
MPhil of Chinese University of Hongkong (Shenzhen)

My research interests include software testing and fuzzing.