Planned 2023-05-04
- Attendees and Apologies
- Previous minutes
- these minutes are from the 2023 April Plenary
- see https://github.com/AGLDWG/meeting-minutes/blob/master/plenary-2023-04-06.md
- Welcoming and introduction to new members
- Follow-up on previous meeting's Action Items
- ToR & MoU status review
- ARDC PID System update
- KG & Spatial
- General business
- Next meeting details
Chair: Nick Car
Scribe: Nick Car, Armin Haller
Attendees: Andrew MacLeod, Anusuriya Devaraju, Armin Haller, Isabel Ceron, John Machin, Joseph Bell, Junrong Yu, Katsman, Nataliya, Kheeran, Lesley Wyborn, Liz Hazelwood, Madhushi Bandara, Margie Smith. Matt Duckham, Michael Henry, Michael Rigby, Mieke Strong, Nicholas Car, Pascal Perez, Phill O'Connor, Pip Bricher, Stephen Pratt, Tina Yang.
Apologies: Len Smith
- New member Liz Hazelwood
- No Action items
- Request to ABS, got our emails, thinking about it, but have not actioned it yet.
- Action Item: Nick to send Michael the MoU and the specifics of the PID service for them to go back to Department of Climate Change, Energy, the Environment and Water and to discuss a response
In readyness for handover:
- PID patterns tidied up ** Refactoring and Syncing was performed, we now see which org has which pattern ** another few weeks of cleaning and moving to remain
- non-critical domain handling removed
- w3id.org
- comprehensive testing re-started
- GeoSPARQL has next version been approved just today. To come out as a new stanard (recommendation).
- Presentation from Nick (link to be shared after)
- Demo and Discussion on GeoSPARQL ** Extension to GeoSPARQL proposed for fuzzy geometry
Postions on Spatial relations in KGs compared to keep them in other systems
- Armin: LLMs/GPT can be used to populate your Spatial KGs automatically from structured/unstructured data sources and will allow users to make human friendly queries that can be translated directly to GeoSPARQL, if a response needs to be computed, i.e., having spatial relations in the KG is important for verification (i.e., autogenerate queries that check the pre-trained data).
- Michael Rigby: Working on a KG of species that exist in a current time in a habitat. Use it to answer queries that things are in context and how to use relations to create a more abstract relations
- Kheeran: How spatial relations can be used in KGs will need to be thought about. How to make use of it in the KG. Concept of a fuzzy boundary is really important and to include that in the KG. How to bring that in.
- Matt Duckham: Fundamental issues of uncertainty in spatial information is not something to be solved by KGs, but the power of KGs is in the ability to linking to other ontologies, e.g., PROV etc.
- Anu: Join of graphs in KGs is important, and linking helps discovering new information which may no be possible with individual graphs
- Pascal Perez: KGs is taking geospatial representations from object-focused to - potentially - agent-compliant paradigm by shifting the priority from the shapes to the features. Then, much easier to link to other ontologies as mentioned by Matt
- LinkedIn operates on a KG: https://thenewstack.io/linkedins-real-time-graph-database-is-liquid/
- Next meeting in June, 2023-06-02