Thesis writing; bottom up vs. top down

Question

Situation: I’m completing a master’s degree in computer science. Currently, I’m at the final stage and finishing writing my thesis.

Usually, similar theses in my institute have the following structure:

1. Introduction.
2. Literature survey.
3. Proposed solution, method, framework, ...
4. Implementation and case studies.
5. Conclusion.

Problem: Actually, the research problem that I’m undertaking is a little complicated and crosses multiple landscapes; it’s about designing data warehouses using ontology. So the way I have followed is bottom-up, providing a chapter of foundations before the solution chapter (point 3 above). These foundations contain multiple basic definitions, constraints and rules that I build upon in the next chapter (including some novel theoretical issues), in order to give the reader the sufficient tools to launch reading my work.

However, my supervisor told me that this is not the best way, and that reader may be confused with all this much of theoretical knowledge in a bottom up fashion, that may lead to getting lost. Instead, he suggested to define any concept only whenever I need to use it, even if it this concept is proposed by me.

Question: Though I know that there may be no specific answer. However, in such a situation, which is better? bottom up or top down? Any suggestions?

Answer 1

I have not enough reputation to comment. So, we will have to go with an answer. I disagree with Cameron Williams and I agree with your supervisor. Your thesis is not a textbook and front loading in a thesis will not have the intended effect. Frankly, if you start with basically a set of definitions, it is quite likely that your assessors will only cursory read that section. They might then be confused when they read the actual solution chapter because they don't have the necessary understanding of the concepts you thought they would have.

I think you should follow your advisor. Think of your thesis as telling a story where you introduce new concepts at the point where you need them. You can always have a glossary at the end of your thesis to gather all relevant terminology in one place.

The exception to this advice would be if you are writing in an area where there is a lot of confusion around terminology and names of concepts. If your goal is to argue that prior literature was sloppy in naming its concepts and you want to clarify the literature, a discussion of definitions at the start could be warranted.

Answer 2

These are just two different styles, and no one can say one is better for all things, though one may tend to work better with certain manuscripts with certain audiences. In this case, since your advisor has specifically suggested top-down as being better, try it his way. Trust your advisor.

It's not clear from the question why you think the top-down approach is not as good. If you're worried that there are too many new definitions etc. that people will have trouble remembering and be difficult to locate, this can be solved with an index or index of notation and/or creating subsections at the appropriate places to treat new concepts in a mini-isolated environment where they're introduced. For readers who already know these concepts, they can just skip those subsections.

Answer 3

It's very common to see

1. Introduction.
2. Literature survey.
3. Contribution 1
   1. Proposed solution, method, framework...
   2. Implementation and case studies.
4. Contribution 2
   1. Proposed solution, method, framework...
   2. Implementation and case studies.
5. Contribution 3
   1. Proposed solution, method, framework...
   2. Implementation and case studies.
6. Conclusion

My own dissertation was originally structured like you have in the question, with all results coming after all framework. This got very negative feedback at my oral defense, and I reorganized it as shown prior to signoff and deposit.

The negative feedback wasn't specifically directed at the structure, but at the fact that I had failed to coherently link everything together, and the chosen structure was a major cause.

In contrast my Powerpoint deck was organized around the three contributions, and was received very positively, which helped me reach the decision to use the same approach for the dissertation.

Answer 4

Ultimately, I think that you should follow your supervisor's advice, unless he/she offers you enough freedom in that regard. However, I would like to emphasize the following aspects:

• Your thesis' structure is very typical for a single topic scholarly works (the rarer alternative, used more frequently in Ph.D. dissertations, is a collection of essays on a common theme) and has a solid research methodology foundation, so I wouldn't worry about that at all, unless your supervisor require you to make structural modifications (in that case, I would politely present my arguments, but comply, if meeting the requirements is demanded upon).

• Terminology-wise, I think that the above-mentioned standard approach to structuring academic work should not be called bottom-up - if you insist on using such terminology, I would rather refer to it as top-down, since you analyze the topic from more general concepts to more detailed (discussion of rationale for choosing between those two approaches is beyond this scope of this answer, but, briefly speaking, I think that it mostly depends on the availability of some theoretical foundation on the topic: if it exists, then I'd use top-down approach, otherwise - bottom-up).

Answer 5

I'd just like to provide a bit of personal perspective. Last week I defended my Masters thesis in Mechanical Engineering. I wrote it in the style you call "bottom up."

My thesis was an experimental comparison of different published literature methods for testing thermoelectrics, so I started with a chapter on thermoelectrics, followed it with a chapter on the testing methods, and then moved to the more traditional goals, methods, results, conclusions for the remaining chapters.

By putting the requisite technical background at the front, I was able to make my discussion of the methods in the experimental section more succinct and discussion of the basic mechanisms mostly out of my conclusions, but I did repeat key concepts throughout the thesis. I think that it is possible to write in a "bottom-up" manner, but remember that you may still have to briefly summarize concepts as they come up in the work.