How to Structure a Conference Paper in Experimental Computer Science

Posted on February 7, 2012 by


Some suggestions by others on the web

read related articles

After my recent posts about titles and abstracts, some folks have asked about the larger task of writing the entire paper.  I haven’t had time to complete that fully, but thought I might share an outline, and some material by others on this topic. First a few links to other’s coverage of this topic:

  • How to write a control paper by  Joa ̃o P. Hespanha  at UCSB, recommended to me by Magnus Egerstedt. I like this one overall, especially he agrees with my view on searchable titles.  But I don’t agree with Hespanha’s assertion that the abstract “should not attempt to justify why the problem is important/timely.”

Send me an email if you think I should add to the above list. Also, here’s a brief outline of how I structure my papers:

Title: See blog post. Bottom line: Search Engine Optimize!

Abstract: See blog post. I believe this is the most important paragraph.

1. Background and Related Work: For conference papers I like to combine these two topics into one section (in lieu of “Introduction”).  Include: What is the problem you are trying to solve? Why is it important? Why is it hard? How have others solved it, including your own previous solutions. When you mention other approaches, suggest reasons your approach is different or better.

2. Approach: What is the key insight or innovation of your approach? If at all possible describe the method simply in words first (not equations!). Then follow up with equations or algorithmic details as necessary. Avoid nitty gritty details of implementation for now.

3. Implementation: This section is not always necessary.  The point is to explain those details not covered above that might vary from implementation to implementation but that others need to know in order to reproduce your results. Also information about the specific implementation that might explain extremely good (or bad) quantitative results: Things like machines you ran it on, language you wrote it in, other details one might need to know.

4. Methodology: Describe the experimental setup, the various conditions of the experiment, how the results were measured and recorded.

5. Experimental Results: Describe in a factual manner, the quantitative results of the experiments. Hold off on rendering an opinion or conclusion here.

6. Discussion and Conclusion: This is were you draw conclusions from the results. You might speculate on the causes of the results or opine on what future work might entail or reveal. Recap what you did, restate the main points from your abstract. Acknowledge those who helped or funded.