This document discusses a hack day event called "ArkHack" where participants worked to link archaeological, cultural heritage, and environmental data. It describes efforts to map environmental data from the SEAD database to linked open data formats. The document then discusses potential uses of linked data from multiple sources to reconstruct past landscapes and environments over time through techniques like paleoenvironmental modeling. Challenges and prerequisites for more fully realizing this vision are also outlined.
Linked Data in Production: Moving Beyond Ontologies
Real-time Visualisation of Cultural Heritage and Environmental Archaeology Data in Landscape Reconstructions
1. Real-time Visualisation of Cultural
Heritage and Environmental Archaeology
Data in Landscape Reconstructions
Marcus Smith, Riksantikvarieämbetet
Philip Buckland, Umeå University
marcus.smith@raa.se
phil.buckland@umu.se
2. ArkHack
• May 2013, hack day hosted at RAÄ in Visby; “ArkHack”
• RAÄ, Kb, SEAD, HUMLab
• Goal: develop information infrastructure for archaeology,
cultural heritage and environment by linking data
• Participants at different stages of linked data:
– Kb (Libris) highly structured, queryable, linked open
RDF from the ground up;
– SEAD also highly structured, but relational, not linked
RDF;
– RAÄ (K-samsök) queryable linked RDF, but coarse
and sometimes inconsistent
3. SEAD data to RDF
• Divided up into teams:
– Link Libris & K-samsök using SPARQL to find objects
in common
– Interface and UX
– Map a subset of SEAD data to RDF
– Vision and use-cases
• Surprisingly successful!
• Stay tuned for more ArkHacks in 2014!
4. How might this information be used…
…beyond simply being present as another
source in K-samsök?
5. Environmental Reconstruction
• Linked data about a
landscape during a
timeslice
• Drawn together from a
variety of sources
• Making complementary
inferences from
different data
• Animating timeslices to
illustrate changes over
time
13. … at and around sites and monuments.
With lots of clever mathematics, programming and GIS!
And qualitative interpretation.
14. Potential Data Sources
• K-samsök
– Sites & monuments from FMIS
– Finds from 40+ national and regional museums
– Fieldwork documentation (not yet!)
• SEAD (not LOD – yet!)
• TORA (doesn’t exist yet, but will be LOD)
– Settlement patterns
• Lantmäteriet (not LOD)
– Historic maps
– National height model
• SGU (machine-readable, but neither linked nor open)
– Geology and bedrock surveys, soils, vegetation, climate
• GBIF – Biodiversity (limited)
• (Some of these data are accessible via INSPIRE Geodata portal)
16. Places a site in a broader context,
• …which is why we want to do it.
• You simply cannot understand a place completely from the modern
landscape. Digital Terrain Models (DTMs) are only part of the story.
• Help understand:
– A more complete and real landscape picture pre/during/post
construction/settlement
– Why they built it/settled there
– What they were thinking
– How they interacted with the environment
17. Further possibilities
• References from structured texts using contemporary
historical sources & literature:
– MENOTA http://menota.org/
– Skaldic Project http://abdn.ac.uk/skaldic/
– IEM http://www.nabohome.org/iem/
• A service that can accept all this data, and spit it out a
beautifully-rendered 3D model.
18. Further possibilities
• References from structured texts using contemporary
historical sources & literature:
– MENOTA http://menota.org/
– Skaldic Project http://abdn.ac.uk/skaldic/
– IEM http://www.nabohome.org/iem/
• A service that can accept all this data, and spit it out a
beautifully-rendered 3D model. (How hard can it be?!)
19. Further possibilities
• References from structured texts using contemporary
historical sources & literature:
– MENOTA http://menota.org/
– Skaldic Project http://abdn.ac.uk/skaldic/
– IEM http://www.nabohome.org/iem/
• A service that can accept all this data, and spit it out a
beautifully-rendered 3D model. (How hard can it be?!)
• Educational systems and games.
20. (How hard can it be?!)
•
•
•
•
A “non-trivial” task…
Different approaches for different proxies and goals
Integrating these into single landscape output…
What is scientifically useful is not often pretty...
Sugita etc.
Bunting, Middleton & Twiddle
21. How far have we come?
Biodiversity, Literature,
Palaeoenvironment,
Cultural Heritage &
27. Challenges
• The data are not linked, services not in place… yet!
• Coverage of paleoenvironmental data is geographically and
chronologically patchy.
• Coverage of proxies is patchy.
• Extrapolation from the “closest” surrounding data points (in time and
space) reduces reliability(?)
• Interpretation is not neutral/objective - depends on models used.
• Visualizing (variability in) accuracy, precision & equifinality.
• Different users prefer different types of visualization.
• Uniting forces of existing computer models for palaeoenvironmental
reconstruction? Multidisciplinary science is difficult and hard to fund.
28. Prerequisites
But much of the rest of the infrastructure for the data itself already
exists:
• FMIS
• SEAD
• K-samsök for managing/aggregating linked data sources
• Lantmäteriet
• SGU
• GEORG (older) and KARL (younger) geometric maps from RA
• Structured historical texts
And others are on the horizon:
• DAP
• TORA
• Excavation database(s)?
ArkHack, goal to develop information infrastructure for archaeology, cultural heritage and environment by linking data.
Some success in translating SEAD data to RDF.
But how might this information be used, beyond simply being present as another source in K-samsök?
But how might this information be used, beyond simply being present as another source in K-samsök?
Archaeology important in complete landscape reconstruction – cannot separate people & nature when reconstructing (or predicting)
Different reconstruction methods vary in complexity, transparency, transferability, flexibility (cf. data quality), comparability (between sites)
Lot of work to do on the archaeological database front...
Despite the fact that the SEAD data isn’t yet linked, and we don’t have a service that can automatically extrapolate ancient climate/environment from an arbitrarily selected dataset, much