Piotr Sowiński

PhD student, AI & semantics aficionado. The crazier the project, the more likely I am to be interested.

[Piotr Sowiński @psowinski_]

What is an #RDF #stream? Everyone seems to have a different idea! In a just-published journal paper, we propose a taxonomy to systematize this and help with streaming system interoperability. @ElectronicsMDPI @MDPIOpenAccess https://t.co/zSztEcc9JW

[Karolina Bogacka @kbogack4]

Experimenting on #ML #inference pipelines in real-life conditions requires not just #performance 🚀, but also #flexibility 🛵. In our newest paper published in @MDPIOpenAccess @ElectronicsMDPI we present a solution. Read more here: https://t.co/pbZ09CQuyC

[Piotr Sowiński @psowinski_]

Do you have an interesting ✨ #RDF dataset? Something big, or funky, or #rdfstar, or live-streamed? Check out RiverBench, the open RDF benchmark suite! https://t.co/v2WLPdsvZd

We'd to have your use case in the suite! 😊
Preprint: https://t.co/ZwChsMtbdU

[Piotr Sowiński @psowinski_]

Does anyone know of any publicly available RDF-star dataset? Something bigger than a toy example :D

[Michael Black @Michael_J_Black]

In the LLM-science discussion, I see a common misconception that science is a thing you do and that writing about it is separate and can be automated. I’ve written over 300 scientific papers and can assure you that science writing can’t be separated from science doing. Why? 1/18

[Piotr Sowiński @psowinski_]

Should we have AI therapists then?

[Piotr Sowiński @psowinski_]

If you are looking for a dystopian story about AI therapists being (unsuccessfully) used to address systemic stress at work, then look no further :( https://t.co/3OUpwAXFEJ

[Piotr Sowiński @psowinski_]

“It’s a joke. It’s trying to gaslight the teacher to say, ‘Oh, this … workload is normal, let’s see how we can reframe our perspective” – that's the dangerous part. A therapist that just plain does not understand you... sounds like a bad idea to me.

[Pavel Klinov 🇪🇺 @klinovp]

A big reason I am excited about correlated subqueries in #SPARQL is that it *requires* defining var substitution. It was underspecced and it created lots of issues with EXISTS + parameterised queries can return diff results with diff vendors. It's actually not straightforward.

[Piotr Sowiński @psowinski_]

@ApacheJena Oh! 😆 Thanks for the tip, I had no idea this was already there.
This will certainly be handy!

[Piotr Sowiński @psowinski_]

An exciting new feature just got merged into @ApacheJena: LATERAL JOIN! 🎉
Great work by @AndySeaborne, good to see Jena improve continuously, it really is a huge driver for my team's research 💖

https://t.co/4dFKC2HIyJ

[Piotr Sowiński @psowinski_]

My (&collaborators') paper on superfast 🌪️ RDF streaming! It's still early work, but the initial results are already great: https://t.co/CbHcgKCPjf

I'm working on a faster, more universal version of it with better tooling 😉 Stay tuned!

[ASSIST-IoT @AssistIot]

#LearnaboutASSISTIoT: Maria Ganzha, our Technical Manager, presented the paper "Efficient RDF Streaming for the Edge-Cloud Continuum” at @IEEEorg 8th WF-IoT, in the context of ASSIST-IoT Special Session.
@NGIoT4eu @HorizonEU

[Piotr Sowiński @psowinski_]

What strikes me most is that this is regarded as "a huge achievement", while the used methods themselves are not breakthrough science. The connections between them are. When you think about it, it's insanely crazy to stitch such systems together.

[Piotr Sowiński @psowinski_]

Great analysis of Meta's Cicero AI! It's a good example of how neurosymbolics & other forms of AI hybridization are crucial to making "real" AI systems. Current SOTA pure-ML AI wouldn't be able to do it ;)
🧵1 / 2

[Gary Marcus @GaryMarcus]

Wondering what @MetaAI's legit impressive new AI Cicero might mean for the next few years in AI?

Hint: It’s not all about scaling.

@ErnestSDavis @NYUScience and I discuss at
https://t.co/6wn8bBUGjv

[Ora Lassila @oralassila]

Me, yesterday, explaining how semantics should travel *with* data. Unfortunately, we instead keep “re-articulating” semantics in every system separately. 📸: @DeniseKGosnell

[Olaf Hartig @olafhartig]

The charter of the RDF-star Working Group has been approved by @w3c! The group, which will be chaired by @oralassila and @linkedktk, is chartered to extend the official RDF and SPARQL standards along the lines of RDF-star and SPARQL-star! 🌟 https://t.co/EnaimhwlU2

[Michaël Benesty @pommedeterre33]

12X faster transformer model, possible?
Yes, with @OpenAI Triton kernels!

We release Kernl, a lib to speedup inference of transformer models. It's very fast (sometimes SOTA), 1 LoC to use, and hackable to match most transformer architectures.
https://t.co/xFK4Yom3gL
🧵

[ASSIST-IoT @AssistIot]

#LearnaboutASSISTIoT: The second day of the ASSIST-IoT 5th GA plenary meeting started earlier today. Partners continue WP presentations and technical discussions in two different rooms.
#ASSISTIoT #IoT #NGIoT #research #meeting #plenary #innovation
@NGIoT4eu @HorizonEU

[Piotr Sowiński @psowinski_]

@SeargeDP fps and rts - something like planetside 2 :D