Florina Piroi
banner
flrp.bsky.social
Florina Piroi
@flrp.bsky.social
TU Wien, Center for Research Data Management, Computer Science, ML/AI. Hobbies are my own.

P.S. I block easily
Reposted by Florina Piroi
Im September 2025 hat das TU Wien Research Data Repository die international anerkannte CoreTrustSeal (CTS)-Zertifizierung erhalten und damit gezeigt, dass die TU Wien beim Aufbau einer verlässlichen und nachhaltigen Forschungsinfrastruktur zu den Vorreiterinnen zählt.
TU Wien Research Data Repository erhält CoreTrustSeal
Die TU Wien gehört dadurch zu den Top-Playern in der Liga der offenen Repositorien.
www.tuwien.at
November 11, 2025 at 7:58 AM
Reposted by Florina Piroi
While cleaning a storage room, our staff found this tape containing #unix v4 from Bell Labs, circa 1973

Apparently no other complete copies are known to exist: https://gunkies.org/wiki/UNIX_Fourth_Edition

We have arranged to deliver it to the Computer History Museum

#retrocomputing
November 6, 2025 at 8:50 PM
Reposted by Florina Piroi
Jede/r fünfte Österreicher:in glaubt an geheime Mächte in der Politik, gleich viele wie in der Schweiz und mehr als in Deutschland (12 %). Eine Studie @unisalzburg.bsky.social zeigt, wie #Verschwörungsglaube und #Politik zusammenhängen.

Mehr auf scilog entdecken 👉
scilog.fwf.ac.at/magazin/euro...
Europas Politik im Bann der Verschwörungen
Im Zuge der Coronapandemie wurden Verschwörungserzählungen zum politisch relevanten Phänomen. Forschende der Universität Salzburg untersuchen in einem internationalen Projekt, wie diese Inhalte in den...
scilog.fwf.ac.at
October 31, 2025 at 9:00 AM
Every time before a lecture starts ...
#teaching #nlp #ws25 #whatamidoinghere
I don’t know what I was going to tell them, but it was going to involve the words “I can’t do my job if none of you fucking listen to me”
October 24, 2025 at 11:36 AM
Reposted by Florina Piroi
FAIR data: 'too little, too late', unless … by @steltenpower:

Some thoughts and suggestions to seriously try for researchers that want to share their data.

https://communities.surf.nl/en/regie-op-data/article/fair-data-too-little-too-late-unless
FAIR data: 'too little, too late', unless …
When people say "make data FAIR" I wonder whether they mean doing extra paperwork after the fact, or changing how data are created so they’re FAIR from the start —
communities.surf.nl
October 23, 2025 at 8:15 AM
Reposted by Florina Piroi
Humans always think they’ve covered their tracks and deleted their data, but they’re wrong a lot.
October 20, 2025 at 8:55 AM
I might want to pin it on all my feeds!
Why the hell does everything have to happen at once?
October 14, 2025 at 12:38 PM
Reposted by Florina Piroi
"'The decision highlights the lack of transparency in #Microsoft 365 Education,' said Felix Mikolasch, a data protection lawyer at Noyb [...] 'It is nearly impossible for schools to inform students, parents and teachers about what is happening with their data.'"

👇 Read the full article 👇
Microsoft violated EU law in handling of kids’ data, Austrian privacy regulator finds
Austria's data protection authority ruled that Microsoft illegally tracked students using its education software by failing to give them access to their data and using cookies without consent.
therecord.media
October 14, 2025 at 5:00 AM
Reposted by Florina Piroi
🎉 Big #WIN for student privacy! 🔒 Austria’s DPA has issued a decision finding that #Microsoft 365 Education illegally tracks pupils and uses student data for Microsoft's own purposes.

📰 Read the full blog post, including a comment by Max #Schrems 👉 noyb.eu/en/noyb-win-...
noyb win: Microsoft 365 Education tracks school children
Favorable decision by the Austrian DSB: Microsoft Education 365 may not track school kids and Microsoft is ordered to provide full access to kids' data.
noyb.eu
October 10, 2025 at 7:38 AM
Reposted by Florina Piroi
🚨 In case you missed it: We've filed a complaint against Whitebridge AI. The company sells “reputation reports” on everyone with an online presence – without a valid legal basis.

More info 👉 https://noyb.eu/en/whitebridgeai-your-personal-data-sale-you-and-anyone
September 30, 2025 at 8:30 AM
Nothing that we didn't know since the emergence of language models. Landmark is only the fact that OpenAI (finally) states it publicly.
September 24, 2025 at 7:18 AM
Reposted by Florina Piroi
Heimische Medien berichten (schon wieder) über eine Schrottstudie eines neoliberalen US-Thinktanks (und verherrlichen ihn dabei als Verbraucherschutzorganisation). Wir und @uebermedien.de haben das schon vor zwei Jahren erklärt: www.moment.at/story/schlec...
Wie österreichische Qualitätsmedien über eine Schrottstudie zu Europas Bahnhöfen berichten - und wem das nützt
Deutsche Bahnhöfe sind schlecht. Wiener Bahnhöfe sind top. Das sagt der European Railway Station Index - ein schwachsinniges Ranking einer neoliberalen US-amerikanischen Lobbygruppe.
www.moment.at
September 16, 2025 at 9:50 AM
I guess it's time to leave this platform, too. What is it whith these CEOs & co. turning up to be nazis in disguise?
Violence is always unacceptable, and threats of violence have no place in public discourse. Our thoughts are with Charlie Kirk and his family.
September 11, 2025 at 9:47 AM
Reposted by Florina Piroi
Was entscheidet, was wir auf Social Media sehen?
Heute bei „Spontan gefragt“ auf KURIER TV: TUW-Forscher @stefanresearch.bsky.social erklärt, wie Algorithmen unsere digitale Realität formen.
🕘 21:15 auf KURIER TV
📺 Mediathek: kurier.tv/spontan-gefr...
July 25, 2025 at 9:11 AM
Reposted by Florina Piroi
Together with www.linkedin.com/in/aaron-cul... I have created a template for making #LLMs (from different vendors and even self hosted ones) easily accessible to researchers - including advanced document RAG with #docling. github.com/complexity-s...
GitHub - complexity-science-hub/llm-in-a-box-template: Template to use genai for chatting and via api to accelerate research
Template to use genai for chatting and via api to accelerate research - complexity-science-hub/llm-in-a-box-template
github.com
July 20, 2025 at 8:21 PM
Reposted by Florina Piroi
AI ist oft nützlich - aber ist sie auch sicher? Mit geometrischen Methoden kann man in manchen Fällen beweisen, dass neuronale Netze bestimmte Fehler garantiert niemals machen werden, zeigt man an der TU Wien:
TU Wien macht Unsicherheit in Künstlicher Intelligenz berechenbar
Mathematische Garantien für neuronale Netzwerke: Ein interdisziplinäres Team der TU Wien präsentiert eine neue Methode bei der Top-Konferenz ICML
www.tuwien.at
June 30, 2025 at 8:36 AM
Reposted by Florina Piroi
🇧🇬🇺🇦In March 2025, a voice on the phone said: “We will bring you to your knees, swine.”
Weeks later, his family was chased on a Bulgarian highway.
🧵This isn’t fiction. It’s the true story of Svet DiNahum - a writer punished for exposing Russian lies.

1/24
May 25, 2025 at 6:23 PM
Reposted by Florina Piroi
Relation Extraction or Pattern Matching? How well do RE models generalise to OOD data? We find that higher in-distribution scores do not necessarily translate to better transferability.

Paper: arxiv.org/abs/2505.12533
May 20, 2025 at 1:47 PM
Reposted by Florina Piroi
Sehr schönes Ö1-Hörbild zum 50. Geburtstag des legendären „Köln Concert“ von Keith Jarrett, über ein verlorenes Klavier, Improvisation und Komposition und die Verwandtschaft zwischen „Köln Concert“ und dem Kitsch-Klassiker „Love Story“:
https://oe1.orf.at/player/20250510/794417/1746860702000
May 11, 2025 at 7:26 AM
Reposted by Florina Piroi
Wanna see something cool?

At the Grammy's in 1976, Ella Fitzgerald and Mel Torme explained what jazz is.

The result? The crowd losing its mind. 🔥
May 4, 2025 at 1:49 AM
Reposted by Florina Piroi
May 4, 2025 at 4:41 PM
Speed killed depth.

Scale killed trust.

The feed killed context.

The result? A media system that can publish everything and still say nothing.
paragraph.com/@signalvs…
April 30, 2025 at 8:55 AM
Reposted by Florina Piroi
I love these silicon valley bros that think ChatGPT has achieved some kind of sentience and therefore might have rights.

Like dudes, you don't even think poor people deserve rights, let's try to achieve that before giving person-hood to the autocomplete algorithm.
April 27, 2025 at 9:24 PM
Reposted by Florina Piroi
April 25, 2025 at 9:50 AM