There is an alternate timeline where the semantic web took off and there was wide investment in ontological tooling to ensure that the information in academic papers, websites, and applications was structured and accessible to future processing.
We instead live in a world where all the useful data is trapped inside proprietary formats, and entangled in meaningless prose - a world primed for large language models to come along and hallucinate the data that might contained therein.
I wanted to actually implement a transformer before I voiced an opinion on large language models. The whole self-attention structure is really cool and highly recommend sitting down and playing with the ideas directly.
I just wish we had done a lot of other work to fully take advantage of them.
@sarahjamielewis
Not just proprietary formats; a lot of academic papers are also locked up by the publishers
Anyone can run their language model on the Wikipedia, only a few researchers manage to get a hard disk from Elsevier
@Madagascar_Sky @sarahjamielewis I found this one painful and I don't agree with all of it, but very well written:
https://twobithistory.org/2018/05/27/semantic-web.html
In my opinion it misses Wikidata, but I'm biased.
@vrandecic @Madagascar_Sky @sarahjamielewis
Okay, right, Denny is biased, he helped make #Wikidata the right way :)
I used to teach how worthless the promises of Semantic Web were. But since Wikidata reached a useful size, there is now quite a lot of useful things one can do with the data, so now I teach that there is a glimmer of hope called Wikidata and federated databases....
@shriramk @WiseWoman @Madagascar_Sky @sarahjamielewis
Those are very relevant observations, and this leads to the one large question the Semantic Web never answered: how does the incentive infrastructure look like? The few parts of the Semantic Web that provided a decent answer to that question were the ones that were successful: schema.org, Wikidata and the wider GLAM world, usage inside emails, inside organizations as a data integration technology.
@vrandecic @shriramk @WiseWoman @Madagascar_Sky @sarahjamielewis
Very interesting debate, I guess one upcoming incentive for semantic tech could be archiving and note keeping. For those that encounter so much info on feeds this could provide better private and collaborative means to collect stuff: ordering, selecting, describing, developing. I am thinking of the experience of using pinterest, tumblr, pinboard, obsidian, etc. as kind of cross-platform extensions to feeds and online collections.
@lukasfx @shriramk @WiseWoman @Madagascar_Sky @sarahjamielewis
Yes, I agree, I think Semantic Web technology is underused in the personal and collaborative note taking space. Hypothes.is is an interesting approach in that direction.
But I'm this space you have to explain what the difference is to bookmarking extensions in your browser, delicious, and the Google Sidewiki, and why it would succeed where these things are not widely used today.
@sarahjamielewis Google has managed to get a bunch of sites to add RDF metadata to their sites in the last few years by calling it JSON-LD.
It seems that telling people it'll increase their search engine performance works pretty well.
I think you describe the problem with the semantic web pretty well. Is not presented in a way people care about.
One should have presented it to researchers as: if you do it, your h-index will increase.
Please don't throw h-index hate my way for saying that. The reason it will increase is pretty simple: easier to use data implies more citations.
Maybe if that had taken off, inventing the appropriate ontological tools would have effectively been == creating AI.
@sarahjamielewis Nice thread you launched there. Back in the day I was heavily involved at W3C and kind of TimBL's loyal opposition, a Semantic-Web skeptic. I still sort of am, but remain open-minded, there's a there there but we haven't found it. In this timeline anyhow.
Having said all that, I object to the phrase “entangled in meaningless prose”. That prose is the real payload, we are language-centric creatures.
Vulgarity and Cloud Orthodoxy in Linked Data Infrastructures - A critical history of the semantic web and linked data, grappling with the next generation of surveillance capitalism, where grand corporate knowledge graphs devour the planet and sell it back to us as a glassy-eyed LLM personal assistants, will we remain stuck in the ideology of the cloud, or can we have better dreams?
theruran 🌐🖥️🖲️
Sarah Jamie Lewis
sabik
Colin Curtin
Madagascar_Sky
Denny Vrandečić
Debora Weber-Wulff
Shriram Krishnamurthi
Lukas Fuchsgruber
jonny (good kind)
James Henstridge
Helge
Mycotropic
Ricardo Segurado
Alvaro
Matěj Cepl 🇪🇺 🇨🇿 🇺🇦
Erwan
Airbag Moments | 45X34 | 🇺🇦
J Low
Vota Shizamura 🌟
WtfPdf
msgbi
RusscamPhoto
Dave Lane 🇳🇿
Tim Bray
Kevin Karhan 
miki