10 steps to start and maintain an evergreen literature survey


“How’s your literature survey coming along?”

I was a first-year graduate student, when a fourth-year Ph.D. student asked me this question at a weekly meeting. I had no clue what a “literature survey” was.

After asking around, I learned that “literature survey” was research speak for “Go find out everything others have done in your field, so that you know which problems are solved and which ones aren’t.”

“Find out everything.” That sounded daunting. How do I know what to look for? Where do I start? When do I stop?

The first step of research is to go down the rabbit-hole of learning about others’ research. You don’t jump into your own research. You look at others’ work first.

These are the things that I wish I had known about a literature survey, about both starting and maintaining one throughout my Ph.D. And about learning to love the process. These are the things I tell my own Ph.D. students today.

Image by Jens Enemark from Pixabay.
  1. Find the top conferences in your field. If you don’t know which ones those are, ask your advisor, and ask other Ph.D. students. When you find those conferences, see if your Ph.D. topic area is listed in the conference’s wish-list of topics for papers. You’ll know you’ve found a good match if your Ph.D. topic area aligns with a topic that the top conferences in your field advertise and seek papers in. For example, in my own area of distributed computing, the top systems conferences are likely to be run by ACM, USENIX and IEEE.
  2. Read the most relevant papers of the last 5 years from those conferences. The last 5 years are often indicative of the focus that the community has. For example, “edge computing” is a term that you would not have seen in systems conferences 10 years ago, but is now a healthy topic. It’s important, as a Ph.D. student, that you know where your field has been heading in the last 5 years.
  3. Read the most-cited papers in those papers. Look up the References section of all of those papers, and find out which papers keep popping up the most across all of them. Read those papers, especially the older ones, as they might be the influential/seminal/must-read papers in the field. You can also use Google Scholar to discover these most-cited papers.
  4. Find the influencers. As you read papers and you see authors popping up more frequently and their papers being cited more often, hunt down the authors’ websites, find out what else they have published, what new results they have, and which other conferences they tend to publish in. Find out if there are any talks or lectures they’ve given, and watch the videos (talks are great ways to learn about key results in a lecture-style format). Put these influential authors down on your “potential external Ph.D. thesis-committee members” list — it’s never too early.
  5. Find your peers. If these authors (in step 4) are faculty, look up these faculty’s Ph.D. students, and see which ones are in your field. Find out in what stage of their Ph.D. work these students are. These are your competitors and your peers in the field. They are walking the same miles as you, but in different shoes. They are the ones who will feel your journey — the highs and the lows — the most. Ultimately, they can also become great collaborators, great sounding-boards, and the best citers of your work. As well as the competitors who “up” your game and spur your work to greater heights.
  6. Write your own summaries of the papers you’ve read. When you write your own summary of a paper, you are forced to capture your interpretation of their work, and that crystallizes it in your mind. Make sure that your summary highlights the novelty, impact, advantages, and disadvantages of each paper.
  7. Categorize papers. Start with a list of keywords, and keep iterating as you read more papers. Suppose that you were asked to organize a workshop around the papers you just read. How would you classify those papers into sessions? Those sessions become your keywords. For example, is this a paper around architectures, algorithms, experimentation, formal proofs, etc.?
  8. Write a living, breathing annotated bibliography document. Take all the paper summaries you’ve written, all of the categorization of the papers you’ve done, and drop them into a document that is never complete until you’re ready to graduate. This can go on to become your Ph.D. thesis’ related-work chapter. And, yes, you can get it published, too. One of my students did just that.
  9. Use a citation manager. All the papers you’ve read may to be discussed in your own papers at some point. You may as well get ready to cite these papers in your own publications. So, start a personal database of references so that you can cite while you write. There are many tools to choose from, Mendeley, Zotero, EndNote, RefWorks, or a simple BibTex file.
  10. Track new publications. Conferences will take place, new papers will be published, new authors will emerge. Set up a Google Scholar alert for key/influential authors in your field, to automate the process of finding new publications.

Rinse and repeat every 3–6 months. You have to keep going down the rabbit-hole throughout your career to satisfy your curiosity and to refresh your literature survey.

“Begin at the beginning”, the King said gravely, “and go on till you come to the end: then stop.” — The King, Alice’s Adventures in Wonderland, Lewis Carroll.

+ There are no comments

Add yours