By: Chamin Morikawa (https://www.linkedin.com/pulse/writing-related-work-section-paperthesis-chamin-morikawa/)
For most students, writing about what they did on their own is not hard. But writing about others' work - which is what you have to do in the "State of the Art" or "Related Work" section - is quite hard for them. Here are a few guidelines to make this task a bit easier.
Let's lay down our assumptions before continuing. I assume that you want to write a "Related Work" section for a research paper or a thesis that describes your approach to solve some problem. Let's also assume that there are other publications that attempt to solve the same problem, but the solutions in them are not perfect. Finally, let's assume that your approach has some difference when compared to those by others, and some improvement (faster, more accurate, easier to afford, etc.).
The question is, how do you come up with a good Related Work section for this publication?
Let's start by looking at the reasons for having this section in a paper or a thesis. While most of you already know them, a reminder can help us to compose it properly.
The primary reason for detailing the state of the art is to highlight that somebody else had not already tried what you did, when you started your research. In order to do this convincingly, you will have to have done a good survey of related research, make a good summary of them if there is a lot of work, and identify the need for improvement. Doing this allows you to demonstrate the motivation for your approach to solve the given problem, and also point out the difference between your approach and the others.
There are a few secondary reasons for writing this section. If you are writing a Master's or PhD thesis, this section serves as evidence of your research skills. Including a good description of the state of the art in a research paper will allow readers who are not very familiar with your topic to learn more about it (if you want your paper to be recommended by professors to their students, this will definitely help). A third reason, one that many researchers won't mention directly, is to have a place for "citations". Citations in other publications is the most important metric for assessing the value of a research publication. A description on others' research can help sustain this metric, and also give an opportunity to make favors (not that I recommend it, but many researchers are guilty of mutual citations and citation loops that boost their records).
One thing to keep in mind when writing the "Related work" section is that it should be shaped "like a funnel". To be more specific, the content should be broad at the start, and focused at the end.
Start with a very brief introduction of the basic research area that your work belongs to. For example, if your paper is about automatic age estimation using digital photos of faces, you can start by mentioning Automated Face Image Analysis as the basic area. You don't have to get down to Computer Vision; Automated Face Image Analysis is already a large research area. Selecting a couple of survey papers to show the advances of this area should be sufficient.
Now it is time to mention other papers that try to solve the same problem as the one that your paper does. You can organize them by idea, into paragraphs, so that the reader won't feel lost among a mix of research works. It is fine to mention the accuracies, and it is essential to mention the state-of-the art if there is a clear evaluation metric to identify it.
Now you are coming to the end of this section. If you are writing a thesis, or a survey paper, this is a good place to summarize the approaches with their performances, advantages and disadvantages, on a table.Otherwise, you can write a paragraph that summarize same content. In either case, the ending paragraph should be a pointer to you work; you point out the limitations in the existing approaches and then state that you are going to try approach X that is different from what has been tried before.
If you are writing a short paper or a demo, you can actually avoid having a specific section detailing related research. The primary reason for this is that the page count for such papers is smaller. In such a case, you can extend the introduction of the paper mentioning work that is closest to your approach. If your solution to the given research problem is very different from others, you can keep the related work section short; but some reviewers might be unhappy with this.
I gave you some guidelines on writing a "Related Work" section, based on my experience as a student, researcher, reviewer, and a teacher. I hope you find them useful. Finally, an unofficial guideline; know the snakes in your jungle! I know of a few reviewers who used to reject papers that did not cite their work, even though their work was so old that they could be cited only in the first couple of paragraphs of the related work section. Your advisor usually knows the culprits for your field.
Helped a lot, thank you!