Deutsch Intern
Robotics

Writing Theses

Frequently Asked Questions

The following guide is adopted from my supervisor Joachim Hertzberg. It was a great help for me.

Format and Size of a Thesis

Theses differ in breadth, depth, and size, depending on the grade they are for, i.e., Bachelor, Master, Diploma, or Doctoral thesis, but their structure is essentially the same:

  • Introduction: Summarize scientific and/or application background, formulate precisely the scientific and/or technical problem that the thesis tackles, state clearly the contribution that the thesis makes (YES, even in the introduction – the reader has a right to know what is in the thesis in order to decide whether s/he considers it worth reading), outline thesis structure.

  • State of the art: Review in necessary depth and breadth the literature, systems etc. that existed before the thesis has started, i.e., the body of scientific knowledge that the thesis sets out to improve at some point.

  • Technical part: The main part of the thesis describing your contribution – its internal structure differs wildly over different theses, depending on what the contribution actually is.

  • Summary, Conclusion, Outlook: The wrap-up part reformulating the contribution, drawing conclusions, if there are any (a summary is not the same thing like a conclusion!), and sketching open issues (only dumb or ingenious theses claim to have solved everything in their ballpark – decide yourself whether you want me to judge your thesis between these two categories!).

 

Unless you have had some prior instruction about how to write a thesis, you should give this matter a bit of thought. There are three things to do here:

  • Ask your instructor.

  • Read prior theses. Check out my backlist of supervised theses. You can download there theses graded with 1.x where the supervisee agreed to publish the thesis online. I have never understood candidates setting out for writing a thesis who tell me they have never before seen other peoples' theses (and don't even seem to find that puzzling). "Humans are case-based" (Roger Schank). Act human-like!

  • Get yourself some guidance. There's tons of stuff out on the Web and on the bookmarket about how to write theses. A good point to start with for an Informatics-related thesis would be Justin Zobel's book.

 

And how long should my thesis be?

There is no fixed rule about that ("If a Bachelor thesis has 90 pages, then it's probably an A" – no!). What matters is the contents and the presentation, where writing concisely is a virtue, but writing too scant means you won't be understood.

As a rough guidance, think of the following sizes as defaults:

  • Bachelor thesis / guided research: 35 -- 60 pages,
  • Master or Diploma thesis: 80 pages,
  • Doctoral thesis: 150 pages,

assuming common-sense formatting. Deviations of these default sizes are welcome if they make sense. For example, if your topic is in image processing, and your thesis contains many images, then it is very likely to be significantly longer. On the other hand, if a bachelor candidate manages to write down the proof that P is unequal to NP in three pages, then he or she will most probably pass.

A point about formatting. You need not use LaTeX. However, if a student/candidate in Informatics delivers a thesis in a layout that is significantly worse than what TeX/LaTeX has been achieving since 35 years now, then s/he needs to explain that to me. (Remember: Science is all about improving on existing results!) This is particularly true for bibliographies that BibTeX would produce for you as a bibliography has to be, without any further effort.

What's in a Thesis Exposé?

When I accept to supervise a thesis (be it a Bachelor, Master, or Doctoral Thesis), I will ask the candidate to produce, as the first milestone in the project, an exposé of his or her thesis. Its size and the time available for producing it differ for the three types of theses: For a Bachelor thesis, it should be available about 2-3 weeks after the formal thesis start; for a Master thesis after 1 month; for a Doctoral thesis after 3 months. However, the idea and structure are equal in all three cases.

A thesis is a one-person research project, and think of the exposé as a project plan. It needs to answer the questions: What is the goal of the project? Where does it start from? Why bother? What is the plan to run it in time? In terms of a research project and a thesis exposé, this leads, more sternly, to the sections

Goal

Describe the goal of your work. This may concern an analytical result (e.g., proving that P=/=NP), or an empirical one (e.g., examine the performance of the 6D SLAM algorithm in an a botanical garden), or a constructive one (e.g., a new algorithm for stereo matching of images taken in complete darkness), or – most frequently for an Informatics thesis – a combination of the three (e.g., a new 3D scan matching algorithm running in O(log log n) and its evaluation in a botanical garden). Normally, the title of the thesis would reflect the goal.

Scientific and/or technological background

Give a sketch of the state of the art that your thesis sets out to improve. In an exposé, the sketch has to be very short and to the point, mentioning exactly the top most relevant papers. For a university thesis, the background includes stating which local equipment and results you will use, if any (e.g., the USARSIM simulator or the real robot).

Approach

If you have absolutely no idea where to start in order to reach your goal, you will probably not make it in time for thesis submission. State here where you will start working. In many cases (typically in Bachelor theses, often in Master theses), a particular approach is enforced as part of the thesis topic that you get.

Expected scientific and/or technological contribution

State briefly in what respect you expect your result to be significant. It should somehow improve on the state of the art, or provide new empirical data, or lead to a result that was never there before.

Work program

Break down your thesis project into smaller steps and make a schedule what you plan to do in which order and in what time. Plan in the order of weeks and months rather than days. If possible and useful, formulate milestones, i.e., important intermediate results. Plan the immediate future in more detail than the distant one. Don't plan for doing all the technical work first, and writing everything down from scratch in the last three weeks: That will almost never work out! Plan to interleave the reading / thinking / programming / experimenting and the writing.

The size and detail of the exposé varies with the type of thesis, according to the calibre of the problem and the available project time. For a Bachelor thesis, think of 3-4 pages; for a Master thesis 10-15 pages; for a doctoral dissertation 15-35 pages.

The author of the exposé is the candidate, i.e., you! Why?: Your thesis supervisors have normally an idea of the thesis topic that they give to you. Your formulation in the exposé shall make sure that you have the same understanding of what you are supposed to work on. Moreover, much of the exposé text may in fact go into your final thesis: a typical introduction shares much of the material with your exposé, and you should have written that yourself.

Be prepared, however, to adjust your topic while working on it! In fact, this is the norm rather than the exception, which leads to the final remark about the exposé: This is a plan for your thesis work, and, like all plans in life, is subject to revision in detail! Don't hesitate to change details of what your exposé says, if it turns out to be necessary. However, do hesitate to change significantly the topic and approach of your thesis that you have described in your exposé – before you do that, consult your instructors!

Reproducibility (a much too unfrequently asked question!)

A valid scientific experiment has to be reproducible. In Informatics, this norm is frequently violated, and even more often, nobody seems to care about reproducing some particular experiment. But that does not invalidate the norm.

For a good or even excellent thesis, this means: If your thesis goals include empirical and/or constructive elements, then your results need to be reproducible for others -- these may be your advisors, and, even better, anybody who reads your thesis and wants to check your results or wants to examine your examples by himself or herself. So, if you write a program or a set of modules as a part of your thesis work, please do make sure that they run at least on our lab computers and that there is somebody of the staff who knows where the software is and how it can be used. Even better, write a little(!) simple(!) interface (which may or may not be graphical) that allows at least your experiments or test runs to be repeated, or, optimally, that gives everybody a chance to do their own experiments or test runs with their own test data. Frequently, bachelor or master candidates, when facing texts from the literature, complain that the reported results cannot be checked or easily varied; and then, the same candidates would often deliver their own results in exactly the same manner that they found so irritating about others' work. Isn't that inconsistent? And, what is worse: It is no sound science!