Write a paper - David's checklist
David's checklist for research papers
- the title is specific. It does not describe a problem area or a class of solutions (unless the class is completely represented in the paper).
- the abstract begins with the contributions. Typically, this means the paper starts, "we present...," or "this paper presents...".
- the abstract summarizes the conclusions (see next point about conclusions)
- the results demonstrate the contributions. If you results do not demonstrate your contributions, then remove them. If nothing demonstrates your contributions, then your paper is probably not ready to submit.
- the paper ends with conclusions. If there are no conclusions, why should anyone read the paper? Conclusions are not just a summary. From this work, what should readers conclude, or what do you conclude? Often the conclusions summarize and emphasize the most important elements in the results and discussion sections.
- there is no "future work." Readers do not really care what you plan to do in the future. If there are interesting open problems to describe, label them as such in the discussion or somewhere else.
- the related work section is not a literature review. A related work section should explain how this work is different from other work. It is not necessary to detail all of the other work, only to describe it enough to explain the relationship to your work. Of course, if you're writing a review article, then this point does not apply.
- the methods section gives only enough detail to reproduce the results. It is not a narrative of how you spent your last N months. It is also not the place to discuss design decisions or funny results. It should very concisely outlined the steps someone else would need to take to reproduce these results.