How to Read a Paper

"How to Read a Paper," Keshav, ACM SIGCOMM Computer Communication Review 2007



I think this paper provide a good way to survey papers.

I use the two steps to survey paper before.

• First, try to read abstract, introduction, and conclusions to see if the idea is interesting to me.
• Second, read the contents carefully.

However, I always get stuck in the second step, and I found the reason after I finish reading this paper. The most problem is that I still not have sufficient knowledge to catch the paper, since I never read the reference. I will try to use the third pass to read paper from now on.

As doing a literature survey, I always go to the top conferences' website to see the recent interesting topics and the relative paper.


I summarize the paper as below.

The Three-pass apporach

• 1st pass
  To read title, abstract, introduction, and conclusions, try to get the overall idea of the paper.
  => Able to answer five Cs : Category, Context, Correctness, Contributions, Clarity.
• 2nd pass
  To grasp the content of the paper, including the related figures and diagrams.
  => Summarize the main idea of the paper, and then go to 3rd pass if in need.
• 3rd pass
  To try to virtually re-implement the paper.
  => The most difficult step, but will be really realize the paper after this pass.

Doing a literature survey

1. Use academic search engine (Ex:Google Scholar) to find some recent papers.
2. Find shared citations and repeated author names.
3. Go to website for some top conferences and look through their recent topics.