Here is a list of common errors encountered when revising paper drafts. Please check these when you write papers (before you send them for me to check).
Write the abstract first. Think of the keywords that you want to include in the title. Then think of a few candidate titles for your paper. Discuss these with your co-authors.
Introduction: List up the things that you want to mention in the intro (same goes for all sections).
- What is the problem?
- Why is it important to solve it?
- What are the challenges?
- How does the proposed method overcome those? (briefly explain the proposed method and give pointers to the corresponding sections in the paper)
- What experiments did you do to evaluate the proposed method? How were the results?
It is a good practice to end the introduction with a summary of the contributions. The Area Chair (AC) will use that to write a meta-review, if they want to accept your paper.
Run a spell-checker
Use a single ~ (tilde) before \cite command such that you have a space between the opening bracket of a citation reference and the remainder of the sentence.
Define all acronyms when you use them the first time. Second time onwards use the acronym you already defined. Acronym package is a handy tool for this.
Check that the font size in your figures are not too small to read. It is a good practice to use the same (or at least similar) font types in your figures as the rest of the paper.
Provide detailed and informative captions to your figures and tables such that the reader could understand what is shown in the figure/table even without having to read the body text of the paper. (bad example: Figure 1: Overview of the method)
Use table, figure labels that are informative. For example,
tbl1is a poor choice for referring to a Table (which might not be the first table in the paper after several edits).
tbl:similarity-resultsis a better naming, which is more informative.
Common word usage mistakes (
If you are performing any statistical significance tests to check whether your proposed method is
significantlyoutperforms a previously proposed method (or a baseline), remember to mention the level of significane (i.e. alpha < 0.01 or 0.05 etc.) (or alternatively the confidence levels corresponding to 99% or 95%) as well as the name of the test you performed. If you do not do any statistical significance tests then do not mention the term
significantin the paper.
If you can spend an hour to revise a section, include a figure, a table etc. that would save 5mins of the reviewer/readers time spent on your paper, then do it!