Center for Analysis and Design of Intelligent Agents

Writing Papers

Purpose of a scientific paper	To communicate knowledge about how the world works efficiently and effectively, through the standard communications style of the chosen scientific field.
Types of scientific papers	There are three main superclasses: 1. Empirical (Experiment) 2. Summary / Review 3. Mathematical
Structure of a typical empirical paper	Abstract Introduction Related work Method Results Conclusions
Abstract	A “mini paper”: A single paragraph that gives the paper's context, the question/challenge, motivation, method, and conclusion in one go.
Introduction	Introduction to the topic, questions and challenge addressed, overview of the motivation and gives its context. The main theory/ies underlying the work. Defines terms to ensure correct interpretation.
Related Work	Overview of what others have done to answer/address the question(s) and why it was not sufficient. Quickly describes how it differs from the work described in the paper.
Questions	Questions to be answered, preferably in the form of hypotheses.
Method	The apparatus, approach, experimental design, etc. in sufficient detail so that someone with scientific training could reproduce the work.
Results	The main answers to the hypotheses posed in the Method section.
Conclusions	A summary of the main conclusions and results. Often includes future work: what the scientists plan to do or recommend shall be done next on the subject.

First 2 Questions: What is My Point & Who Do I Want to Read it?

Ask this before you write your paper	Because a scientific paper has a title and a conclusion, they always have a key point. The answer to these two questions will determine the main message that your paper carries, which in turn determines the experimental paradigm, the methods, the presentation style, and your suggested future work.
What Is My Point?	Your paper has a title which either states your point explicitly or conveys it implicitly. Example title with explicit point: Best-Case Cubesort is Better Than Best-Case Comb Sort. / Evidence for Robot Uprising is Meager at Best. Example title with implicit point: Challenges to Piaget's Theory of Child Development.

Audience: Who Will be Reading My Paper?

Ask this question before you start your research	This will determine your research context, experimental paradigm and the emphasis or slant you choose for your work. This is especially important if you are working in interdisciplinary research or on projects that can appeal to more than one scientific community.
Ask again before you start writing your paper	Select the journal / conference first Do a background search on papers recently published there, to verify that your background section and description of work fits into their context (less important for journals).

Writing An Empirical Investigation / Experimental Paper

Typical Structure of an Empirical Paper

Title	Sufficiently detailed to clearly indicate the main focus, as found in the Contribution part of the paper; sufficiently short to fit in two lines or less.
Abstract	A mini-summary of your paper, intended to allow others to decide whether your work is relevant to their work (and whether they should read on). This section is key!
Introduction	Overall context of the work, short summary of related work and a presentation of the motivation for the work - the problems/questions that are to be addressed.
Related work	Relatively dry discussion and summary of prior work that is relevant to the present work, and how it is inadequate in addressing the problems that your idea addresses, thus necessitating yours.
Questions	Your questions. Your conundrum. This is the heart of the paper. Describe it as clearly as you can.
Method & Execution	The apparatus and setup you used to answer your questions. How you performed the experiment.
Results	What was the outcome of the experiment? Present the results so that they support the claims made throughout - and support the idea that your idea (the topic of the paper) is worth publication.
Discussion	Optional section - sometimes things that didn't fit anywhere else, but really belong in that paper.
Conclusion	This is the conclusion you draw from the work, as presented in the paper. Based on what has been said in this paper, what conclusions can you draw? This is often a semi-summary of the paper.
Acknowledgments	Who sponsored the work; who helped out (but not enough to count as a co-author).
Citations	Related work referred through in the paper.

The Five Key Points in A Paper

What is your topic and why is the topic worth studying?	Present the context and motivation for your work.
What's your contribution?	Scientists are interested in your ideas (the “meat” of your paper). What are you working on? What is your key contribution / idea? Remember, the main emphasis is for the particular paper - do not explain the point of a multi-year research program in a single paper (in any detail), just the point of the material presented in the paper itself.
Why is your contribution important?	To understand your ideas they will need some background (context in the form of motivations, related work). How does your work build on what came before? What does it add?
Can it be believed?	To evaluate and understand your ideas they want to see results of evaluations (results).
Can your results be trusted?	To understand the results you need to explain how you got them (experimental setup).

How the Five Points Map Into The Paper Structure

Topic and motivation	Abstract (1-2 sentences) Introduction
What's your contribution?	Abstract (2-5 sentences) Introduction (briefly) Technology / Topic description
Why is your contribution important?	Abstract (1 sentence) Motivation paragraph/section (often part of introduction, sometimes its own short section)
Can it be believed?	Results section
Can your results be trusted?	Experimental setup / Evaluation sections

Writing Style

Pick your style	- be consistent !
The fewer words the better	Occam's razor works here: As few words as possible, but not fewer (to paraphrase Einstein). A scientific paper must be clear and consistent - there may be no way around being “dry”, compared to e.g. creative writing. Your exciting research subject should make up for it.
Clear sentence structure	Shorter sentences are better than long ones. Shorter = better; longer = worse. Short = good! A paper that is hard to read is a bad paper! Note: When you have written what you think is a really good sentence, there is always a better one that says exactly the same and is shorter.
First person vs. third person	Pick your style - be consistent!
A scientific paper is an argument	A paper presents arguments for a certain state of the world being true. This goes for all papers, including exploratory ones. There is always an argument. Try to make that argument as strong as possible and you will be on your way to a good paper.
A scientific paper tells a story	A story requires that the things described in it are connected: One thing leads to another. The same goes for scientific papers. The human mind has an easier time grasping things that follow logically. If you can't fit everything in the same paper (without making it disconnected or too long) write two papers - or a book.
Acronyms	Avoid them like the plague. Some are necessary, of course (e.g. “IBM”, “NASA”). Don't forget to explain what acronyms mean: “…NASA (National Aeronautics & Space Administration).”

Common Mistakes

Writing to a particular person (e.g. your instructor)	If you are a fiction writer, it may work to write to your mother or lover, but scientific papers are always addressed to a group.
Not following standard templates or guidelines	Most conferences and journals have a standard format and provide templates. Follow the templates!
Formatting the references wrong	Before you decide that your reference style is the most convenient/easiest to read/easiest to set up/best looking, know what conventions you are breaking! To know what conventions you are breaking you must learn the conventions (this can take years).
Not letting the material drive the layout and flow of the paper	If you have answered the question about what your contribution is up front, your material will suggest a certain layout and flow. (Remember, a scientific paper is an argument - it's almost like a lawyer arguing in court.) Try to follow that flow as much as possible. If you try to cram material into a format where it won't fit you will end up with a paper that is difficult to read (i.e. a bad paper).
Not connecting the major points in your paper by a the necessary A-follows-B logic	The only way the human mind can comprehend things is when there is a logical relationship between phenomena and events. Make sure there is a story in your paper.

Reviewing Scientific Papers: Key Roles of a Reviewer

Highlight the paper's strengths	When reviewing, it is always important to note a paper’s strengths, so that the author will not lose these in the process of revision.
The author often forgets the big picture	This happens because it is easy to forget oneself in all the details that have to be right. Point out how the paper could be organized differently to better convey its topic.
Select well what you criticize	Make sure the comments you write are about things that really make a diffierence.
Think like an advisor	Try to turn negative comments into helpful comments.

EOF

Table of Contents

Writing Papers

First 2 Questions: What is My Point & Who Do I Want to Read it?

Audience: Who Will be Reading My Paper?

Writing An Empirical Investigation / Experimental Paper

Typical Structure of an Empirical Paper

The Five Key Points in A Paper

How the Five Points Map Into The Paper Structure

Writing Style

Common Mistakes

Reviewing Scientific Papers: Key Roles of a Reviewer