Enter Note Done
Go to previous page in this tab
Session
  • Presentation | IN31B: AI/ML for Earth Science Datasets, Tooling, and Workflows and Discovery I Poster
  • Poster
  • Bookmark Icon
  • IN31B-0354: PANGAEA-GPT: A Hierarchical Multi-Agent System for Autonomous Discovery in Geoscientific Data Archives
  • Schedule
    Notes
  • Board 0354‚ Hall EFG (Poster Hall)
    NOLA CC
    Set Timezone

Generic 'disconnected' Message
Author(s):
Dmitrii Pantiukhin, Alfred Wegener Institute Helmholtz-Center for Polar and Marine Research Bremerhaven (First Author, Presenting Author)
Boris Shapkin, Alfred Wegener Institute Helmholtz-Center for Polar and Marine Research Bremerhaven
Ivan Kuznetsov, Alfred Wegener Institute Helmholtz-Center for Polar and Marine Research Bremerhaven
Antonia Anna Jost, Alfred Wegener Institute Helmholtz-Center for Polar and Marine Research Bremerhaven
Thomas Jung, Alfred Wegener Institute Helmholtz Centre for Polar and Marine Research
Nikolay Koldunov, Alfred Wegener Institute Helmholtz-Center for Polar and Marine Research Bremerhaven


Scientists studying Earth's climate, oceans, and ecosystems need to work with huge amounts of data stored in online databases. PANGAEA is one of these databases, containing millions of measurements from research expeditions around the world. However, finding the right data and analyzing it can be extremely difficult and time-consuming - like searching for a specific book in a massive library without a proper catalog system.


We created PANGAEA-GPT, an artificial intelligence system that acts like a team of smart research assistants. When scientists type questions in everyday language, such as 'Find ocean temperature data from the Arctic,' the system automatically searches through the database, finds relevant information, analyzes it, and even creates graphs and charts. It's like having several expert helpers who specialize in different areas - one knows about ocean science, another about ecology, and so on.


Our system is particularly good at rediscovering old, forgotten datasets that might contain valuable information about how our planet has changed over time. Early testing shows that tasks that used to take scientists hours or days can now be completed in minutes. This technology could help researchers make important discoveries about climate change and our environment much faster.




Scientific Discipline
Suggested Itineraries
Neighborhood
Type
Main Session
Discussion