Writing papers

Objectives

To publicize an important discovery, and submit it for criticism and improvement.

Let it be Explanatory

Telling a story

Write as if you are writing a story. The abstract and the conclusion should contain mostly the same text, except one may say ’we will show’, and the other should say ’we have shown’.

Don’t try to sound impressive

Academics have the bad habit of using jargon and stilted prose. Explanatory writing sounds less impressive, and is longer. But it does its job better. Otherwise, fewer people will be influenced by the idea, more people will throw it away.

Help the reader!

Keep writing lively, avoid turgidity.

Level of effort

Don’t polish papers endlessly, just go for the deadlines.

Citing others’ work

Ensure that you cite everyone in the program committee relevant to your submission.

If you are writing a grant proposal, ensure that you include a huge number of citations, to indicate to the reviewer that you have sufficient knowledge of the area.

Planning

Itemized plan for each section

Make a plan document for every huge writing project. Write up section names, fill section with itemized list of things which will go in the section.

Useful in collaboration: Can then discuss this outline with others, get their feedback.

Can rearrange content at a high level.

Implementation

Continuity: outline and summarize

In writing paragraphs or articles: state the topic, write the contents, conclude by reminding the reader about what you’ve stated. Always start with an outline.

Sentence = Part which refers to previous sentence(s) + part introducing new idea. Paper, section, paragraph: introduction + point + discussion + emphasis of point.

Repeatedly summarize.

Clarity

Think: “In what possible way can this be misunderstood?”

Avoid using pronouns like this, that and it when there is a gap of a few sentences since prior reference to the subject: instead, be specific.

Terminology

Use consistent terminology - It will help the reader. Stop hunting for synonyms in the thesaurus.

Give names: use ’red black technique’ rather than our technique.

Crispness

Let all sentences have a person-like subject.

Avoid: Weak verbs: “is”, “was”. Avoid adverbs. (Example: extremely important or very redundant.) Avoid double negatives: “Otherwise, you sound not unlike a windbag.”

Formatting

Try to ensure that figures and tables fill the width of the space allotted to them - let them not give the impression of being inserted just to make the work seem substantial.

Always present numbers in tables right-justified.

Figures and tables

In many venues for presenting experimental research (eg KDD), figures and tables can be added to make the work seem more substantial than they actually are.

Presenting research: draw graphs which swoop up: ROC curve.

Dealing with space restrictions

Avoid too much repetition.

Make everything you say substantial. Avoid 2 sentence paragraphs.

Use latex tricks like vspace to reduce space towards the end of the document.

Review

Do a self-review - read it aloud and read it cold.

A week before the deadline, send it out to friends and colleagues, correct bugs.