Are you interested in the lexicon-grammar continuum? So come join us for #LiS2025, organized by the Grad Programs in Linguistics at UFJF and UFMG! LiS features on-site and online courses by the GOATs in the field!
Register at lnkd.in/g4e9jcZT
It’s free!
It’s hybrid!
It’s for everyone!
#everyoneatLiS
Proud to be one of the researchers behind SHADES, the dataset featured in this @wired.com article, alongside @tiagotorrent.com. SHADES tackles bias across cultures and languages, pushing for more inclusive, multilingual AI.
🔗 www.wired.com/story/ai-bia...
#AI #NLP #BiasInAI #MultilingualAI #SHADES
@markturner.org @nicohinrichs.bsky.social @fredbelcavello.bsky.social @viridiano.com
Why are some futures harder to imagine, explain or understand?
We wondered too — so we built a methodology to measure how people think about future scenarios (and how much effort it takes).
🚀 Excited to share that our FutureVision paper was accepted to #CogSci2025!
Curious?👇
arxiv.org/pdf/2502.01597
Fantastic discussion today in my Lang&Cognition grad seminar at UFJF! We kicked off our exploration of how human cognition shapes our thinking about the future with the Decode the Future game by Future Today @amywebb.bsky.social. The seminar is a test flight of FutureMind, my book w/ Mark Turner.
Hello, #AI #NLProc and #Linguistics folks! Come join us for the 6th International FrameNet Workshop in Japan!
More at globalframenet.org/ifnw-2025
Very proud of this collab between FrameNet Brasil and @copla.bsky.social !
Wanna learn more on how FrameNet can help fight #GBV using data from primary care units? My talk at the CWRU Colloquium is now up on YouTube!
youtu.be/lqbycK9Ryho
Ooops! Thanks! Here’s the right link:
markturner.org/cogscicolloq...
Happy to share our research on using FrameNetBrasil to estimate #GBV underreporting at the Cognitive Science Colloquium at Case Western Reserve University. More info on how to join is here:
markturner.org/cogscicolloqS2…
Excelente matéria do @agagomez sobre #viés cultural em #LLMs.
www.uol.com.br/tilt/colunas...
That Japan is the top travel destination in 2025 you already know. But did you know that it is also the place to be if you're are interested in interpretable, responsible #AI, #LLMs and language resources? Well, it is! Join us in March for awesome talks!
globalframenet.org/ifnw-2025
O modelo semântico desenvolvido apresenta desempenho consideravelmente superior tanto em relação a uma IA que só use os dados dos campos fechados do e-SUS quanto em relação a um modelo que usa embeddings de um LLM.
Essa IA rotulou 6 milhões de registros de mulheres atendidas, gerando 25 milhões de pontos de dados anotados para o sentido daquilo que foi registrado nos prontuários. Na sequência, outra IA também desenvolvida no projeto identificou nos dados do e-SUS padrões de violência.
E m parceria com a Vital Strategies, a FrameNet Brasil construiu um modelo semântico baseado em cenas para as áreas da Saúde e da Violência. Depois, construímos um dataset de treinamento anonimizado para uma IA que rotula automaticamente sentenças de prontuários eletrônicos do e-SUS.
Quer saber o que a #Linguística e a #IA podem fazer pelo enfrentamento da #violência contra mulheres? Então dá uma olhada nesta matéria da @folha.com e segue o fio 🧶
Illustrated image of Pranjal Chitale. “Abstracts: A Microsoft Research Podcast” runs along the bottom.
Pranjal Chitale discusses the 2024 NeurIPS work CVQA. Spanning 31 languages & the cultures of 30 countries, this VQA benchmark was created with native speakers & cultural experts to evaluate model performance across diverse linguistic & cultural contexts. www.microsoft.com/en-us/resear...
Hello, world! 🌍
I’ll be using this platform, mainly cross-posting from X and other places
Kicking things off by promoting (to my nonexistent audience 😂) CVQA at NeurIPS!
Oral:
📍 East Meeting Room 1-3
🗓️ Thu, 12 Dec 3:30 pm PST
Poster:
📍 West Ballroom A-D #5110
🗓️ Thu, 12 Dec 4:30 pm PST
Starting now! Join us on Bilibili!
🔗 cognitivescience.hunnu.edu.cn/ICMC2024/hom...
Pre-ordered! Looking forward to reading it.
The most expected read of the year! Thanks once again!
Hello, #linguistics and #NLProc skies!
My EURALEX talk, on how curated databases are key to developing #AI for #healthcare and #multimodality in a culturally grounded, socially responsible way is now available online!
Happy to continue the discussion here!
videolectures.net/videos/eural...
The rise of #LLMs may give the impression that Language Resources are a relic of the past.
In this talk, @tiagotorrent.com shows how curated databases are key to developing #AI for #healthcare and #multimodality in a culturally grounded, socially responsible way.
videolectures.net/videos/eural...
Hello! Oi! Hola!
First time keynoting at an Intelligent Systems conference (BRACIS). So very proud to be talking about our joint project with @vitalstrategies.bsky.social which uses semantic analysis to estimate under reporting of Gender Based Violence in primary care units. 🤩
a correlation between
impolite and offensive discourse towards assistants and growing gender based violence.
Totally… And, again, something lots of people do serious research about (not me), but I have the impression that the medium plays an important role in priming people to behave as if they were giving commands to a machine. I remember this ACL paper where they were investigating +
Not a conversation analyst, but… my wild guess would be a combination of LLM prompting priming with a pinch of “I’m soooo busy and know how to prioritize things so that I crossed politeness and face saving strategies out of my day to day linguistic repertoire…”. Now, seriously… it’s really crazy
👋 hi! I’d love to be added!
The Brazilian Conference on Intelligent Systems kicks off today in Belém! If you’re interested in learning how we use #AI and #Linguistics for fighting Gender-Based Violence in Primary Healthcare, you may like my keynote on Thursday!
bracis.sbc.org.br/2024/
