I recommend the following general strategy for reading and
understanding a research paper:
- You will need to read through the paper more than once.
It almost always takes more than a single read-through of a paper to
really understand and analyze the ideas/work presented.
Here is a suggestion for how to read a research paper:
- First read the abstract, the introduction and the conclusion and look
through the references. If the paper is well written, you should now know what
problem the authors are addressing, why the problem is important, what the
authors solution is, what their solution contributes to the area, and how
the authors demonstrate/prove that their solutions works and that it improves
on other solutions in some way(s).
- Next read through the entire paper starting with the abstract again.
Don't skip over figures, re-read parts that you don't understand.
Take some notes on the paper and write down questions you have as you go along.
Take time to go over
system models, algorithms, proofs, and results to ensure you understand
the work in detail. This is an in-depth read for understanding, however, some
balance is necessary; you may need to come back to some parts
again later to avoid getting too bogged down in one part and lose sight
of understanding the details of the bigger picture. Proofs are one example that
may take a couple in-depth reads to understand.
- Finally, re-read the paper critically. Did the authors do what they said
they were going to do? What are the important ideas? (just because an author
says something is important doesn't necessarily mean it is) Do their results make
sense? Are their methods sound? What assumptions are they making?
How does their work fit in with other similar work?
What improvements/extensions do they contribute?
- Make an outline of the paper's key ideas/contributions
The idea is to create some organized information about the paper that
will help you sort out the details. Highlight the major points of the paper
in as detailed a way as is useful to you.
- What is the problem they are solving?
- Why is it important?
- How does their work improve on or expand previous work?
- What are the key ideas of their solution?
- What are the main ideas of the experimental results and/or proofs that
support their solution?
- What are some future directions/next steps/other applications for this work?
- Create a list of questions:
about parts that you don't understand
about parts where you question their solution/proof/methods/results
- List comparisons of this paper to other related work with which you
are familiar.
To answer some of your questions about the paper, use the related work and
references sections to direct you to other work that may contain important
background to this work. You can also look at papers that reference this
paper to find work that extends or provides a different solution to the
problem this work addresses.
Links to other reading and writing advice