Last week, I went to SWAT4HCLS in Basel and what a good experience it was.
Of course, is the obvious reality that it has been a long, long time since I have been to a scientific conference; I did go to UKCGE workshop on PGR matters last year, but that was one day and was policy rather than science. Definately pre-pandemic, so three years ago or so. I’ve also not managed to get to many SWAT4HCLS because they have often been in December which is not a great time of the year for me to travel.
The conference itself was a simple enough affair: posters, talks and some keynotes. There was a nice mix of material all of which I really enjoyed. The dinner was well done, in a food hall, so you could get what you wanted, with lots of opportunity for discussion. The only thing I can really criticise is the seats in the venue; they were truly awful; it takes real skills to design something some unfit for purpose.
As always, I bumped into lots of people and had some very nice talks about life, the universe and everything. One person I was particularly pleased to meet was Jerven Bolleman who introduced himself at the dinner; we’ve not met before, to either of our knowledge, but we both seen each others names in various places. It’s nice to put a face to a name.
And what are the main things I would take away from the meeting.
First, I would say, SWAT4HCLS stands for “Semantic Web”, which is my first take away: the semantic web is dead, long live the semantic web. The fevered development of standards of the 2000s has gone, the hype has gone, but much of the technology remains and is now in active use, as an equal partner alongside other forms of graph database. The semantic web is mature now and, in exactly the right way, it has become boring.
Second, and following on from the last, the semantic web technology stack doesn’t entirely work and that is still okay. OWL, for example, is layered on top of RDF but doesn’t really gain a lot from this (I think I can be arrogant enough to say that I am one of the few people in the world who really appreciates how difficult the RDF serialization can be, having implemented Horned-OWL). And vice versa, OWL has never become a schema language, with greater rigour than RDF(S). What it has become is the way for building large vocabularies in the form of ontologies and “type checking” them before deployment. Likewise, the “web” part of semantic web hasn’t really worked but the use of URLs has provided us with a good way of producing world unique and sharable identifiers which resolve nicely much of the time. We know what these technologies are used for now, rather than what we hope they will be.
And, finally, lots of people are jumping on the bandwagon of Large Language Models (all scientists are prone to fashion, but computer scientists can beat everyone else in getting distracted by the latest shiny thing). But no one is worried at all about semantics, logical reasoning getting out-competed; we know that we need both to solve the really complicated problems we have in biology and medicine.
Oh, and some meta reflection. Is everything back to normal post-pandemic? Well, I would say we have reached a new normal. The conference was live-streamed, and mostly hydrid; from live side it looked like this mostly worked, as ICBO worked for me last year remote. As an attendee, I enjoyed being able to catch up with people as well as the serendipty of conference; but, I find myself rather more intolerant of the hassles of being away; a 40 minute interconnection in Schipol in one way has never been fun and a four hour connection on return is a different kind of pain; and, how can I avoid mentioning the seats again. I did go, but quite a few people I know didn’t. And, finally, twitter; in the old days there used to be a fair amount of traffic, as well as questions; this time, nothing. For the science community, like FriendFeed before it, twitter is dead.