Conducting Literature Review

Literature review is an essential part of any proper research (and not only) project: it basically answers the simple question of “what is the history of the subject of my project, and what is the current state-of-the-art?”. It aims at helping you understand what are the pitfalls you should avoid, and what are the opportunities for improvement. Through literature review you are researching what others have done in the domain you are working so that you can get inspired about how to work for your own projects. It also helps you build the appropriate language and terminology, so that you can search things more efficiently. So, what are the steps you need to take to conduct and document your literature review?

Understand the jargon - refine your search keywords

Each domain of expertise comes with its own language (often called jargon). Understanding this language is very important as it can help you search resources more efficiently. To understand what the jargon of a domain is, you need to try multiple searches with descriptive queries, until you start getting results that are relevant for you. Then you start looking for words or expressions used frequently in the resources you find. For instance, I once was working on a project where we wanted to make one robot arm pass an object to a second robot arm. We started by looking for articles on this topic, describing the process of ‘passing objects from one robot to another’, ‘multi-robot object handling’, or ‘dual robot object passing’. I don’t remember the exact queries, but hopefully you get the idea. We were not getting enough relevant articles on what we were looking for, until at some point we encountered an article calling this process a ‘handover’. Once we started looking for ‘robot handover’ articles, we discovered a vast number of them on the topic we were interested in. We had just learned the jargon of the domain.

Go for quantity first

Once you have some keywords, go wild and search for as many resources as you can (ok, don’t exagerate). The resources can be research articles (journal or conferences), books, websites, white papers, product pages, patents etc. The type of resources you will find depends on whether your project has a more research or engineering/development component. But the main idea is to try to find many resources from different periods in time (don’t focus only on either too old or too new resources), and from different perspectives if possible. For research articles, you can use google scholar first, as a ‘all-catch’ search engine. For the rest, normal search using any engine will also do. But try to gather enough resources and save them aside for later. You can try to organise them by relevance, type, and age, which will help you going through them later on.

Then filter for quality

Not everything you find on the internet is worth your time reading, so you need to scratch on the surface of your resources and discard the ones that seem low quality. How do you filter and article for quality without reading it all, well that’s something you learn by doing. But you can discuss with your supervisor your findings, and (s)he can help you with the filtering. You might end-up discarding a big part of the resources you initially saved, but that is ok. You need 5-6 really relevant resources that will be the core of your literature review.

Sometimes you will encounter articles which are literature review articles themselves. These are articles that summarise the state-of-the-art in a certain domain, and they can be very useful as they can help you find other relevant articles quickly. However, be careful when using them, as they might be biased towards certain approaches or authors. They can be a good starting point, but you should still try to find the original articles they are citing.

Then you go for depth

The next step is to use your 5-6 resources as the starting point for expanding your research. In each field there are typically a few ‘golden’ articles that many researchers are referring to. So if you managed to find those, then you can search for forward and backward citations. Backward citations are articles that are cited in your ‘golden’ article (therefore older publications). You can find them at the end of the article, and these are the ones used in the introduction or discussion part of your main article. Therefore the chances that they are also relevant are high. Forward citations are articles that are citing your ‘golden’ article (therefore newer publications). These have also high chances of being relevant, but not necessarily. You can find them using tools such as google scholar, by clicking on the ‘Cited by …’ link below the article. The example i give above is for research articles, however similar logic can be used for websites as well. Websites often have links to other pages (similar to backward citations). Finding forward citations can be more difficult though.

Another very interesting search engine specifically for research articles is Semantic Scholar. It has many features that can help you find relevant articles, such as filtering by field of study, publication year, and citation count. It also has a feature called ‘Highly Influential Citations’ which can help you find articles that have had a significant impact on the field. Web of Science and Scopus are other similar platforms, but they are not free to use. However, if you are connected to the university network, you will have access to them.

Then you summarise

So by now you should have 15-20 resources, and the next step is to get the juice out of them and write it down. It should be just some bullet points answering questions such as:

• What methodology did the authors use?
• What are the strong points of the article?
• What are the limitations?
• Do they mention any future work? Typically, this information is found in the methodology and discussion part of the article. The point here is to create a ‘map’ of the articles you found so that you can try to see patterns, pitfalls, and possible future directions for your work.

How to organise your resources

You will start having a lot of articles and resources, so you need to find a way to organise them. There are many reference management tools out there, such as Mendeley, Zotero, EndNote, etc. You can use any of them, as they all have similar functionalities, but I personally use Zotero as it is Open Source. The main idea is to be able to store your articles in one place, tag them with keywords, and write notes/summaries for each of them. This will help you a lot when you will need to write down your literature review, especially if you are going to work with LaTeX. These tools will help you create your bibliography in no time.

Then you write it down

Once you have down the summary, the next step is to write it down in an academic format, as it would appear on your thesis/report. This needs to be part of your thesis/report, so why not write it down already instead of waiting to write it down at the end of your work? Besides completing already one of the most difficult chapters of your thesis/project, you will also gain extra clarity just by the action of structuring it and writing it down.

Final remarks

This is probably the most important part of your work, as it will lay the foundation for the rest of the project. You might be tempted to ‘start working’ on your project as soon as possible, but literature review is work and definitely a very important one. I know that many students are very eager to start coding/building/soldering, but unless you conduct literature review, you will reach a point in your project that you will be wondering what exactly are you trying to do. Of course, you’re not all alone in your quest to conduct literature review. Your supervisor should help you go through the process, and you should discuss with her/him your progress.

Good luck and happy findings!