The Master's
Thesis is your chance to work on something that you decide for a whole
year.
So take it as an opportunity to do that!
And thus, make sure to pick a subject that really interests you.
o Either:
Look at my lists of suggested
projects.
o Or:
Suggest a project of your own in the language processing or computational
creativity fields.
For
some additional project ideas, you might want to have a look at what my
previous students have been working on. You can find most previous theses that
I've supervised at NTNU on NTNU
Open. That will give you some tips of loose ends if you check out the
Future Work sections of those texts. It will also give you a flavour of what
the scope and requirements of a thesis are.
I want to have a chat to you before I sign up to be your
supervisor (e-mail me first so we can agree on a time to meet).
The purpose of the talk is to discuss the projects you are considering, why you
are interested, and what can come out of these projects.
If you cannot come in person (e.g., if you are an exchange-student doing the
fourth year abroad - or NTNU is closed because of a pandemic...), send me an
e-mail and let me know why you have an interest in a specific project, and we
take it from there (probably settling for a Teams meeting instead).
So:
contact me, so we can
discuss the opportunities.
Note that selecting a thesis topic and supervisor is a
3-step process:
1. The
students' deadline 2026 for selecting possible topics is May 21.
2. After
that the supervisors will offer you a (or several) potential project(s).
(No projects will be offered before May 22.)
3. The
students confirm/accept (one of) the offers.
"Shop
around": Select at least 3-5 themes that interest you, from at least
2-3 different supervisors.
Even if you're very interested in working with someone, they might simply get
overloaded...
Also,
be aware that students taking the AI specialisation will normally be
prioritised to get a supervisor in the AI field - this goes for all the staff
working in AI; however, that doesn't mean that I only supervise AI students:
roughly half of my students come from other specialisations. (It also doesn't
mean that all AI-students must have an AI-supervisor.)
Most of my suggested MSc Thesis
topics can be tailored either to one student working alone or to two students
working together (regardless of what the list of
projects indicates).
Working
together may be more fun than going alone, since you may motivate each other
and some of the work (such as exploring the state-of-the-art) must be done
regardless.
However,
be aware that if you team up with someone, both students should have the same
level of ambition and the same goals!
(And that's a lot more important than being good friends from the start.)
Regardless of your grades or previous courses, the most
important thing is that you're interested in the topic you're going to pursue. However,
it will almost certainly be good if you've followed some of the courses where I
am (or have been) involved, such as "Intelligent Text Analytics and
Language Understanding" (TDT4310) or something
similar.
Other
courses that could be relevant depending on the thesis topic include
Information Retrieval (TDT4117),
Recommender Systems/Web Intelligence (TDT4215), and in
general all
AI courses (probably in particular Machine Learning/TDT4173, Deep Learning/IT3030 and/or AI
programming/IT3105).
If
you haven't taken any such courses, you can still do most of my projects (in particular in the Computational Creativity field), but it
may be more work (as you'll have to read up on more material in
order to get to know the field).
[This can also mean taking a course in parallel to your thesis work, including
as one of the theory
modules related to the mentioned courses.]
Normally we'll meet (face-to-face or online) about once
every second week during the year, to ensure that both the work as such and the
report writing is progressing smoothly. At the beginning of each semester, you
should set up a rough time plan for the progress.
Be
aware though that the English term "supervisor" is a bit misleading.
The Norwegian "veileder" and Swedish "handledare" describe the role better: the one who
leads the project is the student, with the teacher being a guide (that's
probably both good and bad news for you: on one hand this entails more freedom
for the student, on the other hand it of course also means more
responsibility).
However, while few students have written a Master's
Thesis previously, I over the years have been involved in well over 130 theses,
so obviously I'll do everything I can to guide you through the process,
including the report writing.
The goal of the first semester is to write a specialisation
report that shows that you're up to speed with the subject area and defines
the actual Master's Thesis topic for the spring
(all project suggestions need to be instantiated - and that's what you'll need
to do during the first semester).
To
do that, you will need to carry out a literature study and write a theoretical
background section. You can also get to know your research field by
implementing a system or running experiments (or reimplementing somebody else's
system/algorithm or rerunning their experiments).
For
the Specialisation Project, the implementation part is not obligatory, but if
you don't implement something, the literature study obviously has to be more substantial.
During the fall semester, you should normally take two theory
modules in parallel with the project work. The two modules together
form a 7.5 credit course, called either TDT4506 or IT3020 depending on your
study programme.
·
You
select which modules you take, but together with your supervisor. You don't
have to select your own supervisor's module, but obviously most of my students
will select either TDT12
- Advanced Language Understanding and Computational Linguistic Creativity or
TDT13
– Large Language Models for Language Understanding and Generation.
Which theory modules are offered a specific year varies (new are added, old are
removed, many are put on hold if the teacher is on sabbatical, etc.; the final
list is normally published in August), but some other modules that can be
relevant (if offered) are TDT04, TDT05, TDT99, TDT55 and TDT17, as well as
TDT70, which is the AI Masters Class on how to work on a Master's Thesis
(writing, topic selection, literature study, etc.), so if you choose that one, the
only theory module which probably isn't relevant is TDT39 (since it covers
similar topics).
Of course
the primary goal of any Master's Thesis Project is to
write a Master's Thesis.
However, a Master's Thesis supervised by me would definitely not be complete unless it included a working
prototype system.
And,
the thesis itself (and hence the project) should be at a level where it would
be feasible to publish the results at an international conference or workshop -
so writing a scientific paper based on the thesis is a clear goal (if
successful, this would normally entail the student traveling to, and presenting
the work at, an international meeting).
Note though that there is no requirement that the thesis work should be
submitted to any conference, only that the quality of the work should be
at that level.
If you have a very keen interest in one of the suggested
thesis topics (or a related one) and even have thoughts about pursuing a PhD in
it, note the possibility to make the Master's Thesis
the starting point of a PhD, financed by the IE faculty.
Björn