/lib

Learn, Imagine, Build
Geoff Messier's Projects & Ideas

How To Do A Literature Search

Becoming aware of the work and people in your research area is one of the most important parts of being an effective researcher. There are two main reasons for this:

  1. It helps you publish. The purpose of the introduction in your thesis or any paper that you write is to provide an overview of the latest research in your area and then to describe how your research addresses some deficiency of hole in the current state-of-the-art. Without comparing and contrasting your work with existing research, it’s impossible to argue that your research is important and new. The danger of failing to reference an important paper is that a reviewer might think that you are not aware of what is going on in your area. This throws the credibility of all of your work in doubt.

  2. It helps you learn. No one person can think of everything. By reading the important papers in your area and following the important groups and conferences, you’ll be exposed to all kinds of new and exciting ideas that you can apply to your own work.

With the convenience of online search tools like IEEE Xplore, Google Scholar and the ACM Digital Library, a literature search seems deceptively easy. It seems like you just need to type in a few keywords, read through the papers that come up and then reference a few of them. While online search tools are extremely important, you must always remember that a good literature search should allow you to answer the following questions:

  1. What are the key, foundational papers that really started off the research in your area?
  2. What are the latest and most important developments in your area?
  3. Who are the people and groups that are doing the best work in your area right now?
  4. What are the conferences and journals where work in your area is most often published?

A good literature search can be divided into two phases, described in the following sections. Phase 1 should be done when you first start your research. Phase 2 should be repeated every month.

To illustrate how these phases work, the following sections will present a case study. “Kevin” is a student who is writing a paper on the aerial deployment of wireless sensors. The paper involves an experiment where he drops real Crossbow wireless nodes from a radio controlled helicopter. Wireless sensor networks is a vast research area so he has decided with his supervisor to focus only on the wireless sensor network research that has an experimental component. He is now proceeding to do his literature search.

Phase 1: Learning your Research Landscape

Step 1: Finding the right keywords.

So, you’ve just started your research project and you’ve established your general area with your supervisor. Now it’s time to do your first online search. The first step is to select the database you should use. A fact of life for electrical engineers is that all the best research is published through the IEEE so doing an IEEE Xplore search is a must. If you are doing network or machine learning related research, you should also do a search on the ACM Digital Library and Google Scholar.

The biggest danger of doing an online search is that you miss the important papers in your area. This happens when you perform a search that is too specific. A rule of thumb is that if you don’t get at least 150 papers coming out of your search, then you need to generalize your search words.

For example, Kevin could type in the search (“aerial wireless sensor deployment”) into IEEE Xplore. This yields no papers. Is Kevin the only person in the world thinking about aerial deployment of sensors? Obviously not. The search was too restrictive. He then types in the following search (wireless AND sensor AND (aerial OR airplane OR helicopter OR UAV OR flight)). This yields 237 hits which is a good number.

The ACM Digital Library search engine is a bit more restrictive. However, an AND search for the terms “wireless”, “sensor” and “aerial” produced 125 hits after isolating for only ACM publications and proceedings.

Some search engines, like Google Scholar, will give you an essentially unlimited list of papers to look at. In these cases, you can look through the first 300 or so.

Step 2: Sorting through the big list.

Do you have to read all 362 papers? Not exactly. Instead, you need to go through the following process:

  1. Read through the titles of all the papers on the list and download PDF’s of the papers with titles that seem relevant. At the end of this process, Kevin has downloaded 43 papers.
  2. Read through the abstracts of all the papers you have downloaded and delete the ones that are not relevant. At the end of this process, Kevin has reduced his paper list to XX papers.

Wow, at this stage you might be thinking that this is going to take you a long time. Having to go through hundreds of titles and abstracts may seem like a lot of work. It is, but you have to do it. This entire process should not take you longer than a few days. While it will be a hard few days, consider that you are going to embark on several years of research based on the information you gather in your search. You risk wasting all of that research effort trying to solve the wrong problems if you don’t start with a good picture of the work going on in your area.

Step 3: Finding the important papers.

Many researchers would stop after Step 2, assuming that their large literature search has netted all the important papers in his/her area. While that might be true in theory, in practice even very general literature searches can miss the really important papers in an area.

In this step, you are going to rely on the literature searches done by the authors of all the papers that you have found so far. Read through the introductions and the list of references of all the papers on your short list from Step 2. After a while, you should notice that many of the same papers will be appearing in reference lists over and over again. These quite often are the foundational papers that all authors consider important. Check and see if you have these papers. If not, download them and add them to your list.

REMEMBER: A good literature search isn’t measured by the number of papers you have in your bibliography (although many of you will end up with very large bibliographies). It’s about referencing the most important papers, not the most papers.

Step 4: Finding the important people.

Behind all research papers are research groups. In every research field, there are a few groups or universities that consistently produce exceptional work. These are the researchers that people make a point of seeing when they attend conferences. Based on the papers you’ve acquired in Steps 2 and 3, identify who these groups are in your area. Visit their university web sites and learn as much about them as possible. You might find papers or information on their websites that is useful to you.

Step 5: Finding the important conferences and journals.

During your literature search, you should also notice a trend in where the papers on your short list are published. There should be only a handful of journals and top conferences dedicated to your area. Identify them because this is where you’ll be trying to publish your work.

Phase 2: Staying on Top of the Research Landscape

After doing all that work in Phase 1, many researchers will often say “Glad that’s all finished!” and proceed with their research work without ever doing another literature search. That’s a huge mistake. New research is constantly being conducted and you need to stay on top of what’s going on in your area. At any time, you should be able to answer the question “What do you think is the most exciting research development in your area in the past 6 months?”.

Fortunately, you can use much of your knowledge gained from Phase 1 to help you do this. In Step 4 of Phase 1, you will have identified the important conferences and journals in your area. Conference proceedings are posted online and journals are usually issued every month. You need to allocate a day or so every month to just scan through the titles of the journals and the table of contents of the conference proceedings once they’re posted online.

As you scan through these titles and papers you should of course be looking for papers that are directly related to your research topic. However, you should also be asking yourself the following general questions:

Some Helpful Tools

When keeping track of the papers relevant to your research area, consider using a paper database program. One example is JabRef which organizes papers and stores references in the native bibtex format used by Latex. However, there are other programs out there.