How to do a master's thesis in my group

Recap. document for a writing a good master’s thesis. May be specific to my field (computer science / software engineering) and my style of supervision.

Concept

A good master’s thesis is a piece of good research.

It is a academic work on a problem of academic relevance, which is anchored in the scientific literature. At KTH, it’s meant to be 6 months of full-time work. The best theses are written by students who are passionate by their topic. Some Master’s theses have changed the world, such as the one on information theory by Claude Shannon at MIT in 1940.

Topics

You already have an idea A thesis is an academic document rooted in the academic literature. What are 3 recent scientific papers closely related to your idea? What would be the novelty of your work compared to them?

You are looking for a thesis topic See https://www.monperrus.net/martin/topics and contact me by email. See also the completed theses in my group: https://www.monperrus.net/martin/past-theses.py.

Novelty, Soundness, Clarity

The goal of a master’s thesis is to contribute to science with a piece of knowledge that is:

• novel: nobody has never done this ever, anywhere in the world. See What it means to be at the state-of-the-art? (Matt Might)

• sound: the logical flow is sound and meets the scientific standards

• clear: the thesis is structured and formulated in a clear way.

Writing an article afterwards: A good master’s thesis contains material for a real peer-reviewed scientific publication, if you’re ready to work the extra mile to make it happen.

Process

Supervisor / Examiner The supervisor is responsible for regular supervision and feedback on both the content and the writing. The examiner is responsible for checking that the thesis meets the expected scientific standards of novelty, soundness, and clarity. Some examiners may also closely follow the work if it is related to their research agenda.

• Tip: select a master’s thesis topic which is directly related to somebody’s research at KTH, you’ll get top supervision.
  
• Tip: PhD students are good supervisors, align your master’s thesis topic with their PhD topics.
  
• Tip: find an examiner who researches or teaches on topics related to yours

Template for Thesis Proposal/Specification The structure must be follows:

• Preliminary thesis title
• Student’s name and e-mail address
• Tentative names of academic supervisor and examiner. Name of industry supervisor if not in the lab.
• Research context:
  • Research area for the project.
  • Problem statement: What is the problem being addressed?
  • Connection to scientific literature: how is the work connected to current research? Discuss and cite at least 3 closely related academic papers. Not all papers are equal, there are obscure conferences and journals which should be avoided, only cite papers that can be found in the ACM / IEEE / Springer / Wiley / Elsevier digital libraries. You can also use https://www.monperrus.net/martin/se-search.html for a clear focus on software engineering.
  • Research questions that will be examined (typically 4, can be 3, see section “research questions” below), with a few sentences for each outlinig the work to be done to answer them.
  • Tentative plan: bullet lists of the initial concrete tasks to be carried out.
• Background of the student: degree / track / specialization
• If in company: where will the practical part of the degree project be carried out, explain what the group/unit does and its composition.

Pilot study / Pre-study The pre-study is the Background + Related work chapter. What goes in “Background” is a introductory presentation of the key concepts and techniques used in the thesis. What goes in “Related work” is a detailed presentation of at least 20 closely related papers, with at least on paragraph per paper. “Background” would not appear in a scientific article (the reader is assumed to know it), “Related work on X” would appear in a scientific article). I suggest to call section “Related Work” on “Related Work on XXXX” to clearly scope the work.

Supervision process

• Progress emails you send a weekly progress report over email to the supervisor and examiner in CC, with three sections with bullet points: DONE, NEXT-STEPS and QUESTIONS.
• Paper summaries: you’ll regularly share paper summaries with your supervisor and examiner.
• In-person meetings can be organized on demand. Feel free to come for a spontaneous conversation in my office
• Feedback is given on a chapter-by-chapter basis
• You are welcome to seat at the lab one or several days per week, with the PhD students, there is a desk for you

Defense 20-25 minutes presentation (strict, fail if not the case).

Professionalism: you are expected to

• work regularly
• give regular progress reports to the stake holders
• warn you supervisor and examiner if you hit a problem (whether technical or personal)
• not postpone the meetings, be on time

Writing

Template: see https://www.kth.se/social/group/examensarbete-vid-cs/page/report-template-2/

Artificial Intelligence (AI)-Generated Text “The use of content generated by artificial intelligence (AI) in [the thesis] (including but not limited to text, figures, images, and code) shall be disclosed in the acknowledgments section. The AI system used shall be identified, and specific sections of the article that use AI-generated content shall be identified and accompanied by a brief explanation regarding the level at which the AI system was used to generate the content.” IEEE Guidelines

Introduction

• (no header) research field, overall topic, context
• Problem Statement (at least two paragraphs: 1. what is the problem you solve? 2. why is it an important problem?)
• Research Questions (one paragraph per RQ, describing the significance and a brief outline of the methodology to answer it)
• Contributions: A bullet list of the contributions to the state-of-the-art, with one novelty explanation per point.
• Outline of Thesis

RQs A master thesis typically contains 3-4 research questions (RQ) that explore different facets of the same problem.

• The research questions must be coherent and related to each other, and give a different light to the same core problem
• The research questions must be neither too vague (because they are unanswerable or the answer would be unsound), nor too specific. They don’t contain vague adjectives and adverbs.
• They are open questions (‘to which extent’,) and not closed questions (which you can answer by yes/no)
• Some RQs are qualitative, others are quantitative.

Case studies Some topics are better evaluated with deep case studies instead of quantitative RQs. All case studies are based on the same methodology.

Bibliography / References: A good thesis contains at least 30 references discussing the related work that: 1) are all relevant to the topic 2) reflect the most recent academic work in the field 3) are discussed appropriately in the text. References could formal bibliographic references or not.

• Formal Bibliographic references

  - Academic papers
  - Mainstream media article with an author and a date
  - Essays or deep technical posts with an author and a date

• Footnote (if the URL is meaningul and looks nice)

  - Tool URLs, documentation pages
  - Tutorials with no author or no date  

• Hyperlink:

  - Link to specific version, commits, transactions, checksums

@misc{example2023,
  author       = {John Doe},
  title        = {An Example Essay},
  year         = {2023},
  howpublished = {\url{https://www.example.com}},
  note         = {Accessed: 2023-10-01}
}

Tips

• You can book a meeting at the KTH Center for Academic Writing for advice about writing.
• Think upfront to the self-assessment report: Problem formulation, Literature study, Method, Results / Evaluation, Discussion (+ Specification, Execution of specification, Autonomy) (+ Sustainability and ethics, Societal aspects)

Good practices:

• A good thesis is factual.
• A good thesis has a good title that gives an idea of ​​its content.
• A good thesis explains a qualified problem and its solution(s). It presents the starting point (the problem) and the arrival point (the solution), but not the tortuous path between the two.
• A good thesis separates the presentation of the problems from that of the solution (the architecture vs. the implementation, the methodology vs. results).
• A good thesis has clear definitions that lay the foundation for the key concepts of work.
• A good thesis is beautiful. Typesetting software makes it easy to produce beautiful thesis (eg LaTeX). The beauty criteria include justification, numbering of sections, links to references, vector graphics.
• A good thesis contains illustrations that ease reading and facilitate understanding.
• A good thesis uses pseudo-code to describe important algorithms. The pseudo-code is at a higher level of abstraction than the implemented code. Code listings are also welcome if they are not too big (<10 lines).
• A good thesis contains hyperlinks that make it easier to navigate to the documents and artifacts mentioned.
• A good thesis has page numbers.
• A good thesis is in an open and well supported format such as PDF.

Bad practices:

• Reusing text without citing it and giving credit to the source (this is called plagiarism)
• Having orphan parts (e.g. only one sub-section, one sub-sub-section).
• Telling the story of what you did (the first day, I did XXX, then, the second day I did YYY).
• Having too much or illegible code.
• Presenting execution traces or unreadable logs.
• Giving too much, too low-level implementation details (variable XX contains YYY information).
• Being vague (“in some cases, XX”). Instead, say “In the case where YYY, XX”.
• Heavy usage of bullet lists (some well-chosen lists may be appropriate but it’s dangerous).
• Spelling, grammatical and typographical mistakes. Sentences that are not sentences (e.g. missing subject, verb or object)
• Heavy usage of bold or italic fonts.
• Using the past tense (the past tense is only allowed in the conclusion).
• Use too many parentheses.
• Have paragraphs consisting of a single line, a single sentence.

See also

• Advice for Successful Master Thesis Projects (Robert Feldt)

• Skills for research

Appendix: KTH regulation

EECS School guidelines: https://intra.kth.se/en/eecs/utbildningsprogram/kursplanering/rutiner-och-regler-for-examensarbete-pa-avancerad-niva-1.1030777

Points taken from KTH Comprehensive guidelines for degree project courses, common goals and assessment criteria

1. demonstrate knowledge of the scientific grounds of their chosen subject area, as well as in-depth insight into current research and development and in-depth knowledge of relevant methodology.

2. demonstrate the ability to search for, gather and integrate knowledge and identify their need for additional knowledge, all with a holistic, critical and systematic approach

3. demonstrate the ability to identify, analyse, assess and handle complex phenomena, questions and situations, even with limited information

4. demonstrate the ability to plan and, with adequate methods, carry out skilled tasks within a given time frame and evaluate this work

5. demonstrate the ability to develop and evaluate products, processes, systems, methods or technical solutions with respect to people’s circumstances and needs, as well as society’s goals in terms of economically, socially and ecologically sustainable development

6. demonstrate the capacity, both orally and in writing, in dialogue with different groups, to clearly account for and discuss their conclusions and the knowledge and arguments on which these are based

7. demonstrate the ability to make assessments with regard to relevant scientific, social and ethical aspects

8. demonstrate the skills required to participate in research and development work or to independently work in other skilled activities