Writing a Research Paper

This is just a very brief introduction to how to writing a scholarly research paper. For more detail, please see the many references online, cited below.

In a nutshell, a scholarly eight-page paper has the following sections:

• Abstract. One paragraph in length. This is the distilled essence of your paper: the problem you're addressing, why it matters, what your solution achieves, and the implications of that achievement. Note that we have skipped idea and especially details here: anything about how it's done. Because this is the part of your paper that you can be sure everyone will read, it's the most important, so it's usually best to write this last. 
• Introduction. About half a page. The problem, followed by your contributions. It's a good idea to make the problem concrete with an example. With contributions, list them clearly, and be specific, so they can be debated (i.e. don't say "It's really cool"). Note that this is yet another summary of your paper, much like the abstract, except offering more detail.
• Problem. About half a page. Now we get into the meat of your paper: this is the most detail you will offer about the problem you're addressing. Many prefer to place their related work section right after this one, when when a discussion of previous solutions to the problem fits naturally.
• Idea. About a page. Here you explain your big idea. Focus on conveying the intuition at this point, not the detail. Use simple examples to begin, generalize later. If you have already discussed related work, this is a natural place to compare your solution to previous solutions.
• Details. About four pages. Flesh out your idea, offering enough detail for others to reproduce your results, and to support your claimed contributions.
• Related work. About a page. Discuss previous solutions to the same problem, and how your solution improves on them. Respect previous solutions; you're standing on their shoulders. Be humble about the weaknesses of your own approach.
• Conclusion. About half a page. Summarize the paper again, focusing on your contributions. Discuss the areas of your current solution that should be addressed, and open problems you think need further work.

A few other pointers:

• You need not use these section titles verbatim, in fact I prefer phrases more specific to the problem context.
• If you are writing a visualization, graphics or human-computer interface paper, pictures are part of your research, and should therefore also be part of your paper.
• Many have taken to using a what I call a "keystone figure", above the title if format allows. This is itself another summary, this time in visual form.
• When explaining your problem and idea, your figures may in fact be simple diagrams, explaining the intuitive examples you use.
• Capture all your images in the highest resolution possible, even if they will be physically small in the paper. This supports zoom into detail in any resulting pdf. Be sure to use lossless compression on your imagery, so that you do not have a completely unwieldy document.
• The IEEE (Viz/Infoviz, others) and ACM (CHI, SIGGRAPH) formats are most common; both accept pdfs generated from either LateX or MS Office, and offer templates for both. Journals often have other unique formats.

Other (more detailed) references on writing research papers:

• Tamara Munzer. Process and pitfalls in writing information visualization research papers.
• Simon Peyton Jones. How to write a great research paper. Perhaps the best insight here is to begin writing your papers before you begin performing the research. He has much more detail on his site, including on good research talks, proposals, and elevator pitches.
• Jim Kajiya. How to get your SIGGRAPH paper rejected.
• Isaacs, E. & Tang, J. Why don’t more non-North-American papers get accepted to CHI?