Following on from my presentation at CAA2014 in Paris, I was invited to submit a paper to a session at CAA2015 covering Linked Data (LD) and focussing on the difference between being theoretically interoperable and interoperating in practice.
The session abstract is as follows:
Linked Data and Semantic Web based approaches to data management have now become commonplace in the field of heritage. So commonplace in fact, that despite frequent mention in digital literature, and a growing familiarity with concepts such as URIs and RDF across the domain, it is starting to see fall off in Computer Science conferences and journals as many of the purely technical issues are seen to be ‘solved’. So is the revolution over? We propose that until the benefits of Linked Data are seen in real interconnections between independent systems it will not properly have begun. This session will discuss the socio-technical challenges required to build a concrete Semantic Web in the heritage sector.
We particularly invite papers that offer practical approaches and experience relating to:
- Interface development and user support for ingestion, annotation and consumption
- Management, publication and sustainability of Linked Data resources
- Building cross and inter-domain Linked Data communities
- Processes for establishing usage conventions of specific terms, vocabularies and ontologies
- Alignment processes for overlapping vocabularies
- Engage non-technical users with adopting semantic technologies
- Licensing and acknowledgment in distributed systems (especially those across multiple legal jurisdictions)
- Incorporation within other software paradigms: TEI, GIS, plain text, imaging software, VR, etc.
- Access implications of integrating open and private content
- Mapping the Field – what components are now properly in place? What remains to be done?
Papers should try to provide evidence of proposed approaches in use across multiple systems wherever possible. Purely theoretical papers and those dealing solely with a single data system are explicitly out of scope for this session.
Keywords: Linked Data, Semantic Web, Web Science
The abstract for the paper is as follows:
Paul Cripps, University of South Wales (firstname.lastname@example.org);
Douglas Tudhope, University of South Wales (email@example.com)
The concept of using geospatial information within Semantic Web and Linked Data environments is not new. For example, geospatial information was very much at the heart of the CRMEH archaeological extension to the CIDOC CRM a decade ago (Cripps et al. 2004) although this was not implemented; a review of the situation regarding geosemantics in 2005 commented “the semantic web is not ready to provide the expressiveness in terms of rules and language for geospatial application” (O’Dea et al. 2005 p.73). It is only recently that Linked Geospatial Data has begun to become a reality through works such as GeoSPARQL (Perry & Herring 2012; Battle & Kolas 2012), a W3C/OGC standard, and the emerging CRMgeo standard (Doerr & Hiebel 2013). This paper presents some real world, practical examples of creating and working with archaeological geosemantic resources using currently available standards and Open Source tools.
The first example demonstrates a lightweight mapping between the CRMEH, CIDOC CRM and GeoSPARQL ontologies using data available from the Archaeology Data Service (ADS) digital archive and Linked Data repository. The second example demonstrates the use of Ordnance Survey (OS) Open Data within a Linked Data resource published via the ADS Linked Data repository. Both examples feature the use of Open Source tools including the STELLAR toolkit, Open Refine, Parliament, OS OpenSpace API and custom components developed and released under open license.
The first example will also be placed in the context of the GSTAR project which is using the approaches described to produce Linked Geospatial Data for research purposes from commonly used platforms for managing archaeological resources within the UK heritage sector. These include the Historic Buildings and Sites and Monuments Record (HBSMR) software from exeGesIS, used by UK Historic Environment Records (HERs), and MODES, used by museums for managing museum collections. As such, the outputs from the GSTAR project have wider applicability in moving geosemantic information from interoperable to interoperating in the UK.
Battle, R. & Kolas, D., 2012. GeoSPARQL: Enabling a Geospatial Semantic Web. Semantic Web Journal, 0(0), pp.1–17.
Cripps, P. et al., 2004. Ontological Modelling of the work of the Centre for Archaeology, Heraklion. Available at: http://cidoc.ics.forth.gr/docs/Ontological_Modelling_Project_Report_ Sep2004.pdf.
Doerr, M. & Hiebel, G., 2013. CRMgeo : Linking the CIDOC CRM to GeoSPARQL through a Spatiotemporal Refinement, Heraklion.
O’Dea, D., Geoghegan, S. & Ekins, C., 2005. Dealing with geospatial information in the semantic web. In AOW ’05 Proceedings of the 2005 Australasian Ontology Workshop – Volume 58. pp. 69–73. Available at: http://dl.acm.org/citation.cfm?id=1151945 [Accessed April 13, 2013].
Perry, M. & Herring, J., 2012. OGC GeoSPARQL – A Geographic Query Language for RDF Data, Available at: http://www.opengis.net/doc/IS/geosparql/1.0.
GSTAR @ CAA: From interoperable to Interoperating
My work on the GSTAR project addresses exactly the issues raised in the session abstract through an investigation of the application of Linked Geospatial Data (LGD) and semantic web techniques for archaeological research purposes; This investigation builds on conceptual structures such as the CIDOC CRM, CRM-EH and GeoSPARQL, incorporating real world archaeological data from a range of sources through to providing working technology demonstrators. This was illustrated through the use of case studies based on my research and also through my work on projects such as the Colonisation of Britain project, undertaken for Wessex Archaeology, and the Later Silbury project, being undertaken for Historic England; the former resulted in a Linked Data resource now online at the Archaeology Data Service and the latter will do too upon completion.
Focussing on producing and then using LGD, my talk looked at the background to my research, the methods, techniques and tools used and some of the pitfalls and successes in creating interoperating LGD resources. I had hoped to be a bit further ahead and be able to demonstrate some map based querying and visualisation in action but at the time, these elements were not ready and interacting with a SPARQL endpoint is hardly the most audience grabbing activity! The research was also put into the context of the broader historic environment sector in England by showing how interoperating geosemantic resources could form the backbone of a digital ecosystem to support research, management and development control functions for a broad range of sectoral user groups.
Since returning from Siena, work has proceeded apace to finalise the geosemantic resource ready for the next phase of activity; taking real world archaeological research questions and expressing these using GeoSPARQL queries to demonstrate the way in which such resources can be used for research purposes. Part of this involves engaging domain experts to see where and how their research interests can be elucidated through the applications of such approaches (more on this here).
Ultimately, the querying and results visualisation components will be housed in a web based interface, hiding the complexity of SPARQL endpoints, to demonstrate how geosemantic resources can underpin user focussed research tools such as Virtual Learning Environments. Whilst it would have been nice to present more of this at CAA2015, the plan is now to complete this phase over the summer ready for a fully fledged demonstration of the whole (completed and submitted) research project at CAA2016 in Oslo.