Reading research papers
Some questions to ask yourself while reading a research paper:
- What problem are they solving? Why is it important?
- What did they really do?
- What is interesting or new about the work?
- What methods are they using?
- Do all the pieces of their work fit together?
- What results did they get? Did they do what they set out to do?
Writing reading reports
Some tips on writing reading reports:
- Answers to the above questions are key in understanding a
research paper.
- A reading report should give a concise summary of the paper,
providing an overview of what they actually did, what methods they
used, and what their results were.
- A reading report should also contain a brief critique of the
paper, telling what you think of the work, what things were unclear or
not addressed, and the merits of the work.
- Pretend that you're writing this report for me, that I haven't
read the paper, and I'm very busy. (Some of these things aren't
pretend!) I am pretty knowledgeable about robotics, so you don't need
to go into detailed explanations, particularly not in the motivations
for the work.
Presenting a research paper
Let me suggest the following format for the paper discussion classes.
This of course is not set in stone and will probably change from week
to week.
- Give a short overview of the paper (5-15 minutes). Introductory
material and related work should be brief. Focus on what they
actually did in the paper, how they did it, and what their results
were.
- At some point, I am likely to jump in. Don't take it personally
or anything. I may just be trying to focus the class more on a topic
that, from reading the reading reports, I thought students didn't
understand or appreciate.
- Since this is a supposed to be a discussion section, not simply a
"students give the lecture" day, we'll let the discussion go free-form
for a while but eventually bring it back to the central issues of the paper.
I view my role in these classes as something of a "co-moderator".
Some tips on presenting a paper:
- Assume that everyone has read it, but perhaps not as thoroughly
as you have. This means you probably won't need to go into much
detail in motivations, but you might for the technical details.
- Have a few topics for discussion. These topics might be
technical details that merit more explanation, interesting issues
raised by the paper, relevant issues not addressed by the paper, or
points regarding the motivation or application of the work.
- Feel free to ask the class for questions or comments.
- Don't get too stressed out about presenting a paper; I'm not just
throwing you to the wolves in the front of the classroom! The
object of these discussion sections is for everyone to develop a
better understanding of these papers. Some research papers can be
hard to digest, and different people will pick up on different
things, so it helps to discuss them with a diverse group.