Your Master’s Thesis – Introduction

Research starts with passion.

IDLab is a research group of UGent
and the innovation hub imec.

• 400+ international Internet experts & data scientists
  • led by Piet Demeester
• 500+ collaborations with industry
• world-class research and education

For your thesis, you’re collaborating
with the KNoWS team of IDLab.

• ±35 researchers
  • led by Pieter Colpaert, Femke Ongenae, Ruben Verborgh
• Our research focuses on
  knowledge graphs on the Web.
  • knowledge graphs
  • Web engineering
  • decentralization
  • Linked Data
  • Open Data
  • …
• Read more about our topics online.
  • Check out my blog posts and our publications.

If we knew what it was we were doing,
it wouldn’t be called research, would it?

Forget about your topic and title.
It was just a way for us to meet.

• We will not tell you what to do.
  • We might tell you what not to do.
• Your first task is to find out
  what it is that you will study.
  • Pick a topic you’re passionate about.
• Read literature relevant to your topic,
  find an interesting topic to work on.
  • We’ll point you in the right direction.

We suggest this research process for
computer science engineering.

1. Study related work
2. Define research questions and hypotheses
3. Build software to test your hypotheses
4. Evaluate your software by running experiments
5. Draw conclusions, and repeat if necessary

Studying the state of the art tells you
what exists and what the challenges are.

• Understand the important technologies.
• Find current contributions.
  • Take note of how they are evaluated.
• Identify the gaps and open problems.

Perhaps the most challenging in research
is to come up with the right questions.

• Design a question that captures your goal.
• Find falsifiable hypotheses you aim to test.
• Research questions can evolve.
  • Break them down into subquestions.
  • Specialize when you find a more pressing problem.

Build software that you and others
can use to test hypotheses.

• Build research software, not production software.
  • Focus on correctness and flexibility.
  • Design modular, interchangeable components.
• Solve the most difficult problems first.
  Avoid spending time on trivial issues.
  • Fail fast and often.
• If you know it’s going to work, it’s not research.

Design an evaluation of your software
that brings evidence for your hypotheses.

• Know before you build what you will measure.
  • Optimize toward a certain goal or target.
• Design repeatable experiments early.
  • Re-run them often to measure improvement.
• All experiments should be reproducible.
  
  • Save your code, setup, and data.

Draw conclusions based on your data.
Generalize, but document limitations.

• Can you accept or reject your hypothesis?
• What does this say about your context?
  • internal validity
• What does this say about other contexts?
  • external validity
• Be specific about what your solution addresses.
  • Report—don’t advertise. Remain objective.

This timeline represents a typical flow.
Yours might differ.

October

send your research questions

plan of action and preliminary steps

November

understanding of related work

December

intermediate presentation

January

first version of code

February

intermediate experimental results

March

improved version of code

April

final experiments

May

thesis handed in

June

thesis defense

The idea owner is responsible for knowing
what you will build and evaluate.

• find the idea
• ask the right questions
• come up with creative solutions

This is you.

The researcher is responsible for
design, planning, and execution.

• design software and experiments
• plan the next steps ahead
• execute the steps correctly and evaluate

This is you.

The project manager ensures that
the research is on track and on time.

• prioritize goals
• fallback to other goals upon failure
• motivate and direct

This is you.

The supervisor helps you check
where you are and where you need to go.

• sounding board
• reality check
• inspire and challenge

These are your thesis supervisors.

The more and the better you work, the more they will.

The advisor will evaluate your work
and encourage you to go beyond.

• mainly interacts with supervisors
• sees your defenses and thesis
• meets you if your work is exceptionally good or bad

This is your thesis advisor.

You can work from home,
but are always welcome at our offices.

• Meet regularly with your supervisors.
  • Meet when you have something to report.
  • Make a clear agenda.
• Feel free to hang around.
  • Creativity comes best spontaneously.

This is your thesis, your opportunity.
Make it the very best you can.

• If you do this right,
  you’ll learn a lot and have fun.
• It’s your opportunity, and your responsibility.
• We challenge you to impress us.

If you like research, and you’re good at it,
we can offer you an exciting career.

• Work on the topics you care about
  with the technology you like most.
• Get paid to educate yourself and
  become a leading expert in your topic.
• Travel the world and spread your message.

Talk to us.
We are here to help you.

• Our primary goal is you doing great research.
  • Evaluation is a means to an end.
• To be able to help, we must know.
  • Report an issue before it becomes a problem.
• Research is in essence an act of communication.