A Critique of the Proposed National Education Policy Reform
CAA 2014: Geosemantic Tools for Archaeological Research
1. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
Geosemantic Tools
for Archaeological
Research (GSTAR)
Paul Cripps
University of South Wales, Trefforest, UK
• Hypermedia Research Unit
• Geographic Information Systems (GIS) Research Unit
Archaeogeomancy, Salisbury, UK
http://gstar.archaeogeomancy.net/
2. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
Introduction
• Background
• GSTAR project
• Case Study
• Conclusions
Earthorama by spdorsey http://flic.kr/p/69C5QD
3. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
CRM EH STAR STELLAR
• CIDOC CRM – Conceptual
Reference Model
• CRM EH – Archaeological
extension to CIDOC CRM
• English Heritage
• STAR – Semantic Technologies
for Archaeological Resources
• English Heritage
• University of South Wales
• STELLAR – Semantic
Technologies Enhancing Links
and Linked Data for
Archaeological Resources
• English Heritage
• University of South Wales
4. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
CRM EH STAR STELLAR
• CRM EH: Extension to the CIDOC CRM
• Archaeological fieldwork data
• Excavation; contexts and stratigraphy
• Finds discovery and processing
• Analysis and Interpretation
• Based on English Heritage Context Recording System
• STAR: Developed infrastructure for working with CRM EH
including demonstrators
• STELLAR: Developed tools for working with STAR outputs
• STELLAR Toolkit
• Inputs: structured data in any schema
• Mapping via built in and user defined templates
• Outputs: Linked Data compliant with CIDOC CRM and CRM EH
5. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
SilburyHillLinkedData
English Heritage, Archaeology Data Service
Linked Data resource built using STELLAR Toolkit
6. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
ColonisationofBritainLinkedData
Wessex Archaeology, Archaeology Data Service
Linked Data resource built using STELLAR Toolkit including Ordnance Survey Open Data
7. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
GSTAR
• Aims & Objectives:
• To incorporate complex geospatial information into our
ontological models
• Vector depictions: lines, polygons
• Investigate advances in geospatial and geosemantic
approaches
• Application of geosemantics and Linked Geospatial Data
approaches to archaeological resources
• Integration of heterogeneous resources via spatial
components of heritage data
• Application of research questions across diverse heritage
resources
8. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
Geospatial Information
• Background:
• Archaeological data is inherently spatial
• Diverse range of spatial information
• Non-spatial data can be related to a spatial component
• CRM EH modelled spatial component using specialisations of
E53 Place
• Stratigraphic Units as Places
• Stratigraphic Units have spatial bounds
• Positive and Negative Stratigraphic Units
• Positive Stratigraphic Units also contain archaeological deposits
• Finds discovered in Places
• Samples taken from archaeological deposits
• Sites, Monuments and Features as Places
• Identified from excavation, remote sensing, surveys etc
9. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
Spatial entities
• Features observed and depicted
• Post Hole
• Stratigraphic units excavated and
recorded
• Fill of Post Hole
• Physical relationships observed
• Temporal relationships inferred
• Interpretation…
• Features grouped and phased to
form Sites and Monuments
• Post Built Structure
• Iron Age Settlement
• Legal Designations
• Scheduled Monuments
10. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
InterpretationofexcavationdatausingprojectGIS
Wessex Archaeology
Spatial component used to aid interpretation through mapping and spatial analysis
11. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
Stratigraphic Units (aka contexts)
• Observed features excavated…
• …recorded and interpreted
• Stratigraphic Unit = atomic unit
of recording
• Product of some (pre)historic
Event
• Deposition of some material
Positive Stratigraphic Unit
• Removal of some material
Negative Stratigraphic Unit
• Recorded on site: descriptions,
classifications, etc
• Surveyed on site: hand drawn
plans, metric survey
12. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
RefiningCRMEHtosupportgeospatialclasses
Positive and Negative Stratigraphic Units
13. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
RefiningCRMEHtosupportgeospatialclasses
Spatial Relationships
14. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
Linked Geospatial Data; integrate
15. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
Story so far…
• Literature Review reveals two converging strands of research
• Two subject areas within distinct domains:
• Semantic Web, Web Science, Linked Data semantics, location,
place, geometry
• GIScience GIS, Spatial Data Infrastructures, web services
• Different approaches
• Different emphasis
• Same ultimate aims
• Working with existing data
• Channel Tunnel Rail Link (CTRL) data from Oxford Archaeology,
archived at the Archaeology Data Service
• Spatial data archived as Shapefile
• Linked Data available as output from STELLAR
16. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
Story so far…
• Case Study: methods for integrating geospatial and semantic
1. Leveraging geosemantic web approaches:
• Triple Stores
• SPARQL and GeoSPARQL
2. Leveraging GIScience approaches:
• Spatial databases
• Web Feature Services (WFS)
• Pros and Cons to each
• Pure geosemantic approach more ‘integrated’
• But very new; ‘bleeding’ edge…
• Performance…?
• Hybrid GIScience approach takes advantage of the best bits of
each
• Stability, robustness
• Requires more complex infrastructure
• More bespoke, less generic
17. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
Geosemantic approach
• All data stored as triples
• RDF, n-triples, XML
• Conversion of geospatial data to RDF
• CRM EH – Context Depiction (EHE0022)
• CRM – Spatial Coordinates (E47)
• GeoSPARQL – E47 as geometry
• Geospatial data included within semantic data
• Well Known Text (WKT) representations of geometries
• Very verbose!
<owl:Class rdf:about="http://purl.org/crmeh#EHE0022_ContextDepiction">
<rdfs:isDefinedBy rdf:resource="http://purl.org/crmeh#CRMEH"/>
<rdfs:subClassOf rdf:resource="http://erlangen-
crm.org/110404/E47_Spatial_Coordinates"/>
<rdfs:label>Context Depiction</rdfs:label>
<rdfs:comment>The Spatial co-ordinates of a Context, defining the actual
spatial extent of the context. Usually recorded at the time of
excavation or other investigative work
</rdfs:comment>
</owl:Class>
18. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
Geosemantic approach
• Context Depiction identifies Context
• <URI for the context depiction> < URI for p87> < URI for the context>
• Context is identified by Context Depiction
• <URI for the context> <URI for p87i> <URI for the context depiction>
• Context Depiction has type EHE0022 Context Depiction
• <URI for the context depiction> < URI for rdf has type> < URI for the
type>
• Context Depiction has geometry
• <URI for the context depiction> < ogc:hasGeometry> < URI for the
geometry>
• Geometry has type WKT Literal
• <URI for the geometry> < ogc:asWKT> < literal = WKT>
20. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
Verbosity…
• Underlying DBMS
restrictions on field sizes
• eg 4K using Oracle
• Extended Data Types
• Implications for
indexing
• WKT can be very verbose
• Complex features;
many nodes
• Precision
• Spatial precision is key
• Not data type precision
• Easy to hit the buffers…
<http://www.opengis.net/def/crs/EPSG/0/27700>"
POLYGON ((569241.09296391497
169487.76102295844,569242.14972065808
169488.63653720432,569242.96691547381
169489.28166943422,569243.69052183023
169489.80626897677,569244.6163534614
169490.31098325696,569245.4501574568
169490.61867947067,569246.84229549242
169491.13088039507,569247.99072701519
169491.45835596515,569249.13880138169
169491.69146708742,569250.70987114881
169491.89583935405,569251.95799411286
169491.83729784336,569251.92909099034
169490.55039028014,569251.44045881392
169490.62534306964,569250.51641985343
169490.64884596801,569249.14214670414
169490.4097076066,569247.53722012346
169490.10687919494,569246.2979823238
169489.74212434812,569245.1707102526
169489.30169979495,569243.96490692452
169488.64508622553,569242.86764335213
169487.49649621252,569242.28527061804
169486.97012858052,569241.09296391497
169487.76102295844))"^^ogc:wktLiteral
21. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
Hybrid approach
• Semantic data stored in triple store
• Geospatial data stored in GIS server
• Geometries accessed via Web Feature Service (WFS)
• Integration achieved by means of WFS URIs
• Java middleware
• Parsing of input queries
• SPARQL queries
• WFS requests
• Parses results
• Uses Geotools libraries for middleware
• Handles WFS requests
• Handles geometries and geometry collections
• Same outputs as geosemantic queries
• But leveraging ‘traditional’ GIS spatial functions
22. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
Stack
• Oracle VirtualBox
• Virtual machine
• Oracle 12c Spatial &
Graph
• Spatial database, triple
store
• Jena
• Semantic framework
• Oracle Weblogic
• Web server
• Geotools
• Java GIS toolkit
• Stellar Toolkit
• Structured data Linked
Data
• Geoserver
• GIS server
• Eclipse+Maven +JDK
• Java programming IDE
• ArcGIS, QGIS
• Spatial data management
• Gruff
• Visualisation of Linked
Data
• Built on AllegroGraph
23. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
Conclusions
• Emerging geosemantic approaches most suitable for
integration of geospatial and semantic data
• Standards compliant structures
• Standards compliant query mechanisms
• GeoSPARQL can be integrated within ontologies and Linked
Data resources
• Simple solution as presented, based solely on W3C/OGC
standards
• More complex & powerful solutions using CIDOC CRM +
GeoSPARQL (eg Heiber & Doerr: CRMgeo)
• Some issues eg precision, structures, verbosity
• Rapid development in this field
• New systems/platforms emerging
• Continual improvements; eg imminent next version of
GeoSPARQL
24. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
Acknowledgements
• Thanks to:
• University of South Wales – funding, supervision, advice
• Archaeology Data Service – data from their archives
• Wessex Archaeology – data, photographs and images
• Wiltshire Council – access to the Historic Environment Record
(HER) data
• Wiltshire Museums – access to museum collections data
• Personal thanks
• Supervisors/Advisors: Doug Tudhope, Mark Ware, Alex Lohfink
• Research group: Ceri Binding, Andreas Vlachidis, Keith May
• Peers and colleagues: Michael Charno, Chris Brayne, Ant Beck,
Sarah May, Gerald Heibel, David Dawson
• Image Credit
• Earthorama by spdorsey http://flic.kr/p/69C5QD
25. GSTAR – Computer Application & Quantitative Methods in Archaeology 2014 – Paris, April 2014
fin
• paul.cripps@southwales.ac.uk
• paul@archaeogeomancy.net
• @pauljcripps
• gstar.archaeogeomancy.net
• hypermedia.research.southwales.ac.uk
• gis.research.southwales.ac.uk