SPM87 | Supporting the Earth Intelligence paradigm of GEO: The EIFFEL experience (by invitation only)
Supporting the Earth Intelligence paradigm of GEO: The EIFFEL experience (by invitation only)
Convener: Dimitris Bliziotis
Mon, 15 Apr, 14:00–15:45 (CEST)
 
Room 2.43
Mon, 14:00
The Group on Earth Observations (GEO) is currently formulating its post-2025 Implementation Plan to serve its new Strategy. The paradigm of Earth Intelligence is prominent therein, among others calling for integration of earth, social sciences and new technologies (incl. Artificial Intelligence-AI, Machine Learning-ML etc), combining different scales and sectors and with special emphasis on user needs. Despite the fact that AI is expected to further thrust the design of services that provide Earth intelligence, there are also other ways to facilitate the use and experience of EO. The meeting aims at presenting what was implemented in this direction over the past 3 years by the H2020 EIFFEL project, how the different aspects of AI came into play and to discuss potential AI routes for the GEO community to pursue over the next years.

The first part of the session will comprise brief presentations on the EIFFEL pilots which utilized GEO infrastructure, such as GEOSS Platform (GEO DAB+ GEO Portal), GEOSS datasets and socio-economic data in unison with local authorities to build diverse climate change applications. These applications culminated in GEO Knowledge Hub entries, available to a much wider pool of potential users. One of the objectives is to identify how this process, i.e. an H2020 project contributing to the GKH, can be mainstreamed.

The second part revolves around the different aspects of AI touched upon during EIFFEL. A Large Language Model (LLM) was fine-tuned based on a compiled, expansive corpus, and utilised for the GEO Infrastructure search and metadata augmentation. The developments of the GEOSS Platform Plus (GPP) and EIFFEL projects based on this work will be presented. The training corpus was further utilised by the Knowledge Centre on Earth Observation (KCEO) to enhance semantic search in a variety of databases (including CORDIS and Copernicus). Further AI aspects include of course data processing but also explainable AI (XAI), a cornerstone in creating trustworthy information. The second, and major objective, will therefore be to map these AI elements against the GEO Strategy’s Earth Intelligence paradigm and respond to the growing needs coming from the GEO community.

Participants, including members from EIFFEL, GPP, GEO, EuroGEO and the KCEO will discuss how the above know-how can accommodate the rapidly evolving AI landscape to the benefit of GEO and Earth Intelligence.

Agenda:

-Introduction, scope (5mins)
-EIFFEL climate change Pilots, using the GEO infrastructure in unison with local stakeholders (8mins)
-GKH, paving the way for other Horizon projects (8mins)
-The EIFFEL Large Language model and the GPP (8mins)
-LLM co-benefits and other AI aspects (8mins)
-Open Discussion including (60mins):
• Mainstreaming GKH for Horizon projects.
• A generalised “GEO” LLM.
• Articulate a first response to the GEO AI foreseen activities.