The structure of paper/report in Systems

Michalis Faloutsos
U.C. Riverside


Some rules:
Recursiveness 1: In every section, the first paragraph should be a summary/overview of the section.
                         You can actually start by saying: "In this section, we present..."
Recursiveness 2: In every paragraph, the first (or second line) should correspond to summary of the
                        paragraph.
Top-Down:      Start from high level and start increasing the details slowly.
Prepare the reader: Reports are not suspense thrillers. The reader should have an idea of where you are
                    trying to get at.
Adapt:              These rules are meant as guidelines. Paper requirements can differ. However, when you
                         break one of these rules, be aware of it and have a good justification.
Personal pet-peeve: huge introductions: if you can not say it in 1.5 page probably it is not very clear in your mind
                               or you don't use the right level of abstraction.

Note: I have probably broken all these rules. I usually ended up paying for it.

Total Paper length: ~15 pages.
Not all sections are compulsory from the list: Our Idea, Analysis, Experiments, Discussion.



Abstract
[1 paragraph 200-300 word. This is the ad and trailer for the paper: it has to be an executive summary of your
work in such a way as to attract attention and get the reader to read more.
Think of how you would explain the work to your office mate]

1-2 lines: goal of this paper
1-2 lines: motivation and importance of work
1-2 lines: per result or contribution of our work
               [it is good if you can quantify your claims with numbers that
                refer to intuitive or easily explainable metrics i.e. 50% faster]
 

[At paragraph level: * indicates one or at most two paragraphs
                               ** means as many paragraphs area needed, but don't do crazy]

Introduction [1-1.5 pages]
* What is the problem, motivation, importance, main contribution
* The problem: high level definition, significance
* Previous work [at most one line per relevant paper, leave details for next section]
* Our contribution: possibly list of results and contributions
* "The rest of this paper is structured as follows. In section 2,....."

Background and Model [1-1.5 pages]
* Definitions, scenario, model
* Previous work [provide details only if this will help show where your work is different or better.
                            Make sure you reference all related work and especially that by people in the committee!]
* Weaknesses of previous work or the absence of the thing you are proposing
* Assumptions and limitations of your work. [Be honest, but don't condemn your work].

Our  Idea [as required]
* The innovation in brief
* New definitions, if necessary
** What it is and how it works
** Explanation of fine points, clarifications

Analysis [as required]
* Overview of what we prove in this section
* Analysis-specific definitions [i.e. a graph structure that you will use in the proof]
** Analysis - proofs
* Comments - Meaning of proofs at an intuitive or practical  level

Experimental Results [ as required]
* Overview of experiments: what we do measure, what we don't, why
* Simulation model or measuring infrastructure
** Experimental Results
        - use a heading (in bold) to distinguish experiments (i.e. The effect of size on performance   or even
            better The size does not affect the performance..)
        - each plot should have a clear reason for appearing (ie with more load we get worse throughput)
        - explain each plot: the axis, what we see, what is the trend, why this is the trend
        - statistical comments will strengthen your results: confidence interval, correlation coefficient etc.
        - Each figure should be fairly stand alone and self explanatory: captions should be readable and understandable.
        - Comments, discussion, explanations

Discussion [1-2 pages]
[If you have space and ideas, you can strengthen the importance of your work by
  showing: a) significance, b) relation to other pieces of work, c) possible practical applications]
* Overview of section
** Paragraphs according to needs

Conclusions [1-1.5]
[Conclusions are very important. Do not expect that the reader remembers everything you told him/her.
Having stated the definitions, you can now be more specific that  in the introduction]
* Overview what this work was about.
* Main results and contributions
* Comments on importance or
* Tips for practical use [how your results or experience can help someone in practice or
    another researcher to use your simulator or avoid pitfalls]
* Future work. Reinforce the importance of work, but avoid giving out your ideas].

References
[It has to be correct, complete, and accurate]

Appendices [as required]
[Appendices are no man's land: you can write whatever you want: reviewers are not obliged to read
through them.  Suggested uses:
a) provide additional results that strengthen your arguments but are not crucial
b) maintain the flow of the paper by putting here a lengthy proof or discuss a detail
c) address an issue that shows you have thought the issue in depth