690 likes | 704 Views
Scientific Communication CITS7200. Lecture 6 Writing a Paper. Writing is part of the process of research. If you don’t publish it, you haven’t done it!.
E N D
Scientific Communication CITS7200 Lecture 6 Writing a Paper
Writing is part of the process of research. • If you don’t publish it, you haven’t done it!
“There are three necessary steps in useful research; the first to begin it, the second to end it, and the third to publish it.”Michael Faraday
But… • No one wants to read your paper. • Concise and relevant title • Inviting abstract • Good visual organisation
Organisation • Target your audience • Choose appropriate language and notation • Rank your contributions • Determine the length • Aim to be concise • Avoid repetition • Achieve general results that give special cases
Science vs. fiction • In both cases, all necessary info is given • But… • In fiction, clues are hidden • In scientific writing, clues are up-front
Hiding clues • State your facts in a way that seems clearly to be doing so for another purpose He was proud of his old college associations, and wore his college tie and gold tie-pin to all formal events. (The victim was stabbed with a small sharp object)
Place your clue somewhere is a long list of irrelevant facts The autopsy revealed that the victim had a high level of alcohol in the blood system at the time of death. There were head wounds from a fall, numerous pin-prick needle marks on the inner arm, buttocks and one on the neck, bruises on the legs, and an enormous tattoo across the back claiming allegiance to a motorbike gang.
Placement of clues • If observers are asked to watch out for something, they seldom remember what immediately preceded or followed that something In the autopsy report, what followed the pin-prick mark on the neck?
In scientific writing, clues must be presented clearly, together, up front, and in a way that makes the deduction that is drawn from them completely obvious.
Abstraction hierarchy of a paper Title Abstract Introduction Conclusion Other Chapters
Title • Single most important part of paper • Directly determines the number of readers • Most readers don’t get past the title
The title is at the highest level of abstraction • It is a concise description of the paper • It captures the content in one phrase • It must attract the attention of the casual browser
Linear Scheduling is Nearly Optimal • Learning the Unlearnable • An application of the multiedit-condensing technique to the reference selection problem in a print recognition system • A Frame-based Computer Vision System • An Initial View on Size Estimation for Expert System Applications • Active Intelligent Vision using the Dynamic Generalised Hough Transform • RAPID - a video rate object tracker
Authors • Author list, with affiliations • Always use same appellation • Use complete mailing address, including email
Lyndon While School of Computer Science & Software Engineering The University of Western Australia 35 Stirling Highway, Crawley, W. A. 6009 lyndon@csse.uwa.edu.au
Always date your work • Establishes precedence • Keeps track of drafts • On a line by itself after the author field
Abstract • Second highest level of abstraction • Summarises the content of your paper • Can be used as stand-alone • 200-600 words in length • Usually a single paragraph • Contains no reference pointers, no abstract symbols
Should be easy to understand • Should be easy to translate • Must lay claim to new results • Avoid starting with “This paper…”
Keywords • Give a list of keywords • Used in computer searches
CR classification • Computing Reviews Classification System • Four-level tree • Three numbered levels and an unnumbered level • Give at least the top two levels for everything you write
D is for Software • D.3 for Programming Languages • D.3.3 is Language Constructs and Features • D.3.3 [Software]: Programming Languages - Abstract data types.
Introduction • Third highest level of abstraction • Must state your thesis or hypothesis
A thesis is a proposition laid down or stated as a theme to be discussed and proved, or to be maintained against attack. “Shakespeare was a republican” “The piano-movers problem is NP-complete”
A hypothesis is a supposition or conjecture put forth to account for known facts, and it serves as a starting point for further investigation by which it may be proved or disproved “Internet usage is growing at an exponential rate”
Explain the problem • Explain what your hypothesis is for solving the problem • Outline your plan of attack
Your first sentence must carry some essential information, but at the same time gently introduce the reader to the rest of the article.
“Finite-state verification tools deduce properties of finite-state models of computer systems.” • “Network protocols such as routing protocols are difficult to test because meaningful experiments may involve dozens or even thousands of hosts and routers.”
Standard body • Introduction • Literature Review • Methodology • Experimental Results • Conclusion
Hierarchy of standard body • Introduction • Conclusion • Other chapters
Literature review • Almost all scientific work is done in context • The lit review explains the context of your work • It reviews the science • Must use consistent notation and level of abstraction
Methodology • Your work uses existing methods • Frequency domain techniques in signal processing • Genetic algorithms in optimisation • MPEG compression in video • AODV routing protocol
You develop new methods • Explain how your ideas work
Often in computer science you develop a new algorithm • Not enough to list the steps • Need to address correctness and complexity
Give some or all of • The steps that make up the algorithm.
The input and output, and the internal data structures used by the algorithm.
The scope of application of the algorithm and its limitations.
The properties that will allow demonstration of correctness, such as preconditions, postconditions, and loop invariants.
A complexity analysis, for both space and time requirements.
Use standard mathematical notation, not programming specific notation e.g. xi rather than x[i]
Use standard mathematical operators • ==, • a = b = c, • a++, • for (i=0; i<n; i++)
Must give enough detail so that the reader can implement the algorithm • Don’t use pseudocode, use English • Good practice to make source code available through your website
Experiments • Many papers describe computational experiments
The scientific procedure • Hypothesise • Test • Refine your model • Repeat
Experiments are done to • Gain insight • Compare existing models • Verify theoretical predictions • Tune algorithm to code parameters • Measure performance
Experiments need to be designed • Identify clearly your hypothesis • Explain how the experiment tests it • Explain your results • Readers must be able to reproduce your experiments