How to find a research problem in a haystack of papers

Comb through the haystack in a methodical way.

Your professor hands you a haystack of research papers, and asks you to go “find a research problem so you can work on it.”
How do you get started?

Read one paper at a time

Pick a paper, and try to get into the heads of the authors.

Step 1: Read to find what was done.

As you read the paper, write down the answers to these questions.

What did the authors go after?
What do they claim is new in this paper?
What do they say was done before them? (hint: Related Work section)
What are the big a-ha insights in this paper? (hint: Conclusion section)
What surprises did the authors encounter along the way?

Step 2: Read to find what’s left to be done.

Read the paper again, now searching for words like “limitations,” “assumptions,” and “lessons learned.”

Assumptions represent the sandbox in which the authors played. You and your work can make that sandbox bigger. As you look at each assumption the authors made, ask yourself, “Why is that assumption there? What happens if that assumption is no longer true?” Boom. Maybe your next paper.

Limitations represent the range of applicability of the results in the paper. As you read about each limitation, ask yourself, “Is that limitation excessive? How do I make the limitation go away?” Boom. Maybe another paper.

Lessons learned represent the challenges the authors ran up against, and how they overcame them (or not). These lessons can be illuminating as they can indicate what challenges still remain, and perhaps what challenges you might run into.

Image by The Digital Artist on Pixabay.

Step 3: Read to find how to do it.

Read the paper again, now searching for words like “implementation,” “experiments,” “testbed,” “real-world system,” “data collection,” “user studies,” “data set,” etc. Understand how the authors went about their work.

Did they build a system? (hint: Approach or Methodology section)
What kind of examples did they use in the paper?
Are these real examples or hypothetical ones?
Did they run experiments in the lab? (hint: Experiments section)
Did they talk to real users? (hint: User Studies section)
Did they do mathematical proofs?
Did they run field trials?
Did they use data from other sources?
How did they analyze the data? (hint: Analysis/Algorithms section)
How did they choose to present their findings? Graphs? Tables?

How the authors went about their business can tell you how to go about yours. It can tell you what kinds of experiments to run, how many experiments, what examples to focus on, how to collect data, how to analyze data, how to present data, etc.

It’s also a good opportunity to think about picking a lane — reading these papers can help you determine what you would enjoy the most. Do you want to be a system builder, an experimental scientist, a theorem prover, a data analyst, or a user-studies expert? Or, perhaps a bit of everything?

Reading can also help you arrive at your own writing taste and style. Learning how authors write up their results is a good opportunity to learn how to write up yours. Make note of the elements of style that you liked — section titles, format, types of graphs, types of images, the transitions between sections, the overall organization of the paper — so that you can use the best of these ideas in writing your own paper.

Step 4: Read to hunt down more to read.

Head to the References section of the paper. Your expanded reading list is right there. Look up some of these paper titles in Google Scholar or other online research repositories, to see how well-cited they are. Focus on the more recently published papers, and add them to your haystack.

Rinse and repeat.

So, you’re done reading. What now?

At this point, you’ve combed through the haystack of the papers that you were handed. You even made the haystack bigger by adding more papers to read. And, you took notes along the way.

You’re now in a fantastic position to write a summary of:

  • All of the solved problems,
  • All of the unsolved problems,
  • The different approaches to solving these problems, and
  • The different styles of presenting the solutions.

You are now sitting on a pile of juicy research problems, multiple approaches to go after them, and multiple styles to write about the results.

You just have to pick the problem, the approach, and the style.

It’s in your hands now.

+ There are no comments

Add yours