User Tools

Site Tools


faculty:denton:writingpapers

This is an old revision of the document!


Writing Papers (Dr. Denton)

Discussion on what to put into a paper. Also some notes on deadlines for the next papers.

Alan's Paper Writing Recommendations

  • Use present tense
  • Avoid short paragraphs
  • Punctuate equations

Paper Components

Abstract, Introduction, and Conclusion are fairly fixed. The others are required but may be arranged as appropriate

 I. Abstract
        * Summary of the entire paper.  Not an introduction, so maybe one line of motivation here.
 II. Introduction
        * Motivation section...
        1.Why do people need it?
        2.How is it new?
        3.Why/how is it successful?
 III. Background
    1. Related Works
    2. Background formalism (e.g., ARM definitions)
 IV. Algorithm
    * Should first concentrate on results, then define the algorithms with really new math (should be about 2 pages)  [This is a good milestone]
 V. Results
    * Plots- Good place to start the writing process (after motivation)
    * Additional experimental description can wait.
 VI. Conclusion

2. How is it new? Can be answered with:

  • More accurate (be careful, this is tempting but hard to achieve)
  • Better scaling/complexity
  • Solves a new problem
  • A new perspective to an old problem

3. How is it successful? Can be shown by:

  • Results from 2
  • Better numbers for accuracy and complexity
  • New approach is better than naive approach
    • Naive must be practically naive (for completely new approaches can be very much so)
    • New solutions may only be able to compare to the latest old solution

PLOTS: Probably need about as many as half the number of pages. For ICDM the pages is 12, so need about 6 figures.

  • One to show Quality
  • One to show Efficiency
  • One or two for partial results (performance key concepts)
  • One or two drawings of key concepts (such as graphical version of data or search space)

Be sure that figures are legible at the size they will be printed/read at. This means legends should be correct (readable) size, text is generally kept at the in-text size, and series lines should be distinguishable.

As Michael Stonebraker has said we need Respectable Graphs and Equations (RGE)

  • Graphs are as above to show results
  • Equations are the new math in the algorithm section
faculty/denton/writingpapers.1484350755.txt.gz · Last modified: by localadmin