Can't decide what to place first, literature review or theoretical description? (Master Thesis in Computer Science and AI)

Question

I am writing my master thesis in Computer Science and AI. I am very undecided about what to put on the first place, and changed my mind many times already. Because of that, I keep rewriting the contents to fit the structure.

I have read from multiple sources that literature review must come first, then the theoretical description, however it feels wrong to me. Because how could I talk about a series of works that involve many hard and specific concepts without first explaining them first? If I have to assume that the reader is familiar with the concept, why should I have a theoretical description then? Or am I misunderstanding the purpose of these sections?

Answer 1

The answer is partially based on the prevalent academic culture. Some places have very concise ideas how a thesis should be structured and some are much more free. In some places, it is some bureaucrat who checks your thesis for conformance to a given standard, in some places it is the thesis director, who might or might not have a certain scheme in mind, and in some places, anything that the committee agrees with is fine.

From the later perspective, a theoretical perspective in general, followed by a literature review, followed by a theoretical perspective more geared towards the problem at hand would be a nice enough scheme. The purpose of the literature review is to place your work in the context of the current state of the art, but also to show that you read what you were supposed to read (and understand). If the literature review comes first, it is more than just an elaborated list of authors - result pairs, but includes putting the thesis in context, which is very similar to a "theoretical description".

Ultimately, a master thesis has usually a very limited audience for which you are writing. Most theses will be read just by the members of the committee or the rapporteurs, or whatever it is called. There are memorable exceptions to this such as a thesis that gets converted into a book or a famous paper, but these are still numerically exceptions. If possible, address your question to them.

TLTR: Not all theses follow a boiler plate and a generic scheme is not a good idea. However, academic customs vary, and some places insist on a given scheme. It is not worth your time and effort to fight that. But maybe you can ask your advisor.

Answer 2

For a master's thesis, you should ask your advisor and do whatever they tell you to do. There are so many different standards and expectations that, practically, what it comes down to is that whatever the person who is going to approve or critique your thesis is the best approach to go with.

This is not a matter of "the advisor is always right"; rather, it is that there are many good choices and it comes down to preference rather than a single or even merely a best "right way" to do things. When it is a matter of preference, then it is always a good idea to go along with your advisor's preference to keep them happy.

However, if as you indicated in comments, it is not easy to get your advisor's immediate response on this question and you want to advance with your work, then I recommend that you carefully observe the structure of the articles that you are reading and then follow the patterns that strike you as clearest and most logical. You will likely find that there are many different patterns in practice, but as long as you follow a pattern that is already published, then you have some precedent to justify the pattern that makes the most sense to you. Moreover, if your advisor later questions your choice, you will have published work to refer to as a justification for your choice; your advisor would be more likely to consider your choice seriously and might even let you go ahead with your preference because of the precedent that you refer them to.