Mathijs' Research Stay: Developing Temporal Reasoning at Oxford University
Why not? That was my guiding principle throughout all of my research stay.
It was - quite literally - the very beginning of this research stay. A day of tinkering with MeTeoR,
a reasoning engine developed at Oxford University, ended at the couch with my girlfriend.
Thinking aloud, I uttered the words "Imagine I would do a research stay there. That would be cool,"
which was followed by a "Why not?" from my girlfriend. Fast forwards across some merge request,
online meetings and an impromptu trip to Montpellier, the plans began to crystalize and four months later, I was boarding the Eurostar towards Oxford.
I got to work with Przemyslaw Walega, a post-doc at
Oxford's Computer Science Department
and assistant professor at Queen Mary University in London. His team did quite a lot of work on DatalogMTL,
and that's what our research for this stay was all about.
Together, we worked out the idea to embed
the temporal reasoning of DatalogMTL into existing non-temporal reasoners.
This idea - when proven successful - will allow us to leverage existing reasoning engines,
instead of having to develop temporal reasoning engines from scratch. This was quite an ambitious undertaking,
but fitting for a stay at one of the world's highest-ranking universities.
On a day-to-day basis, I spent my time at Oxford's Computer Science department.
That in itself was already noteworthy. It's hard to put the allure of Oxford - its buildings, its history,
its atmosphere - into words. The first that comes to mind is "stately".
Even though I've seen my fair share of old and historic buildings,
Oxford's are impressive not only in quantity, but also in quality. They are so well-maintained,
it's hard to believe they are as old as they are. It's a sight that never bores and never ceases to amaze,
not even after months of cycling through them every single day. For a change of scenery,
every now and then I travelled to London, to Queen Mary University,
soaking up the vibe of a younger university and to experience the bustle of the big city life.
It provided for some interesting opportunities such as some local seminars by local professors and companies.
The work we did was quite intense. Along the way, we realized the problems we wanted to tackle
were considerably more difficult than initially assumed.
Nonetheless, we were able to push through, learning quite a lot about the strengths and
limitations of our approach in the process.
Not only did we succeed in creating a sound and correct theoretical mapping from DatalogMTL into regular Datalog,
we also managed to test our theory with several different reasoning engines.
The work was quite intense, but not without result:
we managed to bundle our work just in time for submission for
the AAAI conference on Artificial Intelligence.
And following that work, we're already discussing potential next steps with parties eager to hear what
we did and what we can do with it, both inside and outside academia.
Luckily, it wasn't all work. I joined the local judo club, met great new people and
got to be a guide for familiar faces coming over to visit me from time to time.
3,5 months is a surprisingly long time, but the new experiences,
the skills I gained and the connections I made made it absolutely worth it.
I would wholeheartedly recommend my fellow PhD student to seek out and do for a research stay.
Just ask yourselves: "Why not?" and embrace the adventure.