1. Semantic Digital Humanities Workshop
Oxford, September 2015
The Platform in Use
Dr. Barry Norton
Development Manager, ResearchSpace
British Museum, London, UK
14. ResearchSpace
• The ResearchSpace project:
– is funded by the Andrew Mellon Foundation;
– develops a set of cultural heritage research tools;
– uses Metaphacts platform as a basis to reuse and
combine these tools –
• for each project using ResearchSpace,
• for applications beyond cultural heritage;
– configures and specialises these tools for data
integrated using the CIDOC CRM ontology.
15. ResearchSpace Graph
• Resources are typed into classes in rich poly-
hierarchies:
object/YCA62958 crm:E22_Man-Made_Object
rdf:type
crm:E19_Physical_Object
crm:E24_Physical_Man-Made_Thing
crm:E18_Physical_Thing
crm:E71_Man-Made_Thing
rdfs:subClassOf
16. ResearchSpace Graph
• Different structural parts are represented as
separate resources:
object/YCA62958
crm:P65_shows_visual_item
object/YCA62958/inscription/1
object/YCA62958/inscription/2
object/YCA62958/inscription/3
crm:E34_Inscription
rdf:type
17. ResearchSpace Graph
• The same separation of identity exists for non-
tangible resources:
object/YCA62958
crm:P108i_was_produced_by
object/YCA62958/production
crm:P24i_changed_ownership_through
object/YCA62958/acquisition
crm:P30i_custody_transferred_through
18. Navigating the Graph
• Relation between simple entities (object,
place) become complex to navigate:
crm:P108i_was_produced_by
?
?
?
?
?
19. Querying the Graph
• SPARQL queries are beyond most users:
SELECT ?object ?place
{?object (crm:P46i_forms_part_of |
crm:P148i_is_component_of)* /
crm:P92i_was_brought_into_existence_by /
^crm:P9_consists_of? /
crm:P7_took_place_at /
crm:P88i_forms_part_of* ?place}
20. Aim
• Provide a search-based system that allows
users to navigate relationships more
abstractly:
– Abstract nodes into 6 broad fundamental categories (FCs);
– Abstract paths into fundamental relationships (FRs)
between FCs;
– Allow conjunctive search over FRs on a single FC domain;
– Allow reuse of search definition/results (via FR) in new
search;
– Explain each search result in terms of underlying graph.
21. Fundamental Categories
Things
• Objects
• Parts
• Digital
objects
• Articles
• Letters
Actors
• People
• Org.s
Places
• Towns
• Countries
• Sites
Times
• Dates
• Date
ranges
Events
• Named
events
• Prod.s /
Mod.s
• Periods
Concepts
• Terms
29. 29
Toumba
GRAVITATE Background
Slide reused from Karageorghis, Kiely, Hermon et al. ‘Towards an integrated cross-disciplinary approach to the study of Salamis Terracotta Statues’
31. Basic Search for GRAVITATE objects
• Objects can be found using places, actors
(people and groups) and terminologies
– Currently terminologies direct from Merlin
32. Type of Salamis statuary
• Many shards of terracotta figures with some
recognisable features (morphology)
33. Description of statuary
From text:
• Material: terracotta (known)
• Object type: statue (known)
– Modifier: over-lifesized
• Morphology: hand & wrist
– Modifiers:
• Right hand
• + Clenched
35. Morphology model
• ResearchSpace must also:
– link individual parts to
(annotations over) 3D
model
– link these to an overall
(complete) body
morphology
• but also allow assertions
over sets of parts:
– “these are all similar
beards”
– “these are all parts of the
same statue”
38. 38
A new color, previously un-
observed - yellow - has been
identified inside the earrings
decoration.
Materials analysis
Slide reused from Karageorghis, Kiely, Hermon et al. ‘Towards an integrated cross-disciplinary approach to the study of Salamis Terracotta Statues’
39. 39
Techniques analysis
Vertical marks
Superimposition of incisions
C2412, front side
Various styles and coincidences of
incision not currently recorded,
but provide clues to families / re-
assembly.
Slide reused from Karageorghis, Kiely, Hermon et al. ‘Towards an integrated cross-disciplinary approach to the study of Salamis Terracotta Statues’
40. Challenges
• How to attach scientific/geometric analysis to
individual shards (BM to document model for
GRAVITATE)
• How to enrich search using these richer
descriptions
• How to assert over sets:
– families
– re-assemblies (possibly cross-collection)
• How to enrich search over these sets