Tuesday, May 10, 2005

Reviewing Paper

Currently, I am reviewing two papers for a conference, can't tell you which one - reviewer confidentiality :-), which has an acceptance rate of only 20%.
This means, statistically speaking, I should reject both the papers. Now, if all reviewers (each given two papers) thought and did exactly like me, the conference would have to be cancelled. Ya, right! You forget something called the 'program committee' . They are good at throwing dice.

Jokes apart, reviewing a paper - specially if you are not in a grudging mood (reject wholesale) - is a difficult job. Real good papers cause no problems - strong accept. Same for the pathetic ones - very strong reject. But, what to do about those with 'incremental improvement over state-of-the-art, under certain assumptions'. The problem is that most (both papers on my desk) fall under that category.

Let's say 10% of the papers are extremely good. Another 30% extremely poor. So, out of 10 submissions, 1 is strong accept, 3 are straight reject. We still have to choose one out of the remaining six. These 60% cause a headache. You see, you cant be just too harsh on the paper you got - to the joy of those you didn't.

Anyways, for the reader, assuming you are a first year PhD, starting to oil your publishing engine, here are a some quick feedback on increasing your chances. First, choose examples which CLEARLY explain your contribution. Too simple, the reviewer thinks you are dumb. Too difficult, he just skips it (to come back later to...). If the research-content is poor (half-baked) don't waste time (topping the cake) polishing the paper with diagrams and 'Shakespearean English'. You might as well spend that time improving the quality of your research. The polishing is not a 'in lieu of', but is a must over and above the good research. One of the papers I am reviewing has 50 references. They occupy 1.5 pages out of the total 6 pages. Wow! What a homage to previous art.


Post a Comment

<< Home