Presentation given by Claudia Marinica of theUniversity of Cergy-Pontoise at the ARIADNE winter school about work on supporting Semantic Interoperability in the Conservation-Restoration Domain in the PARCOURS project. The presentation discusses the challenges of achieving interoperability between the various databases existing within the domain and the work that is being carried out to build an ontology based system for data integration.
Claudia Marinica - Supporting Semantic Interoperability in Conservation-Restoration Domain: The PARCOURS project
1. Supporting Semantic Interoperability in
Conservation-Restoration Domain:
The PARCOURS project (2013-now)
Claudia Marinica1 and Cheikh Niang1, 2
1ETIS Lab - ENSEA / University of Cergy-Pontoise / CNRS 8051
2Research Laboratory for Historical Monuments / LRMH - CRC - USR 3224
Claudia.Marinica@u-cergy.fr
2. Ø C2RMF (French Museum’s Research and Restoration Center)
§ Research, restoration and archiving for museums’ artifacts
Ø ETIS Lab – ENSEA/University of Cergy-Pontoise/CNRS 8051 (Equipes
Traitement de l'Information et Systèmes)
§ Computer Science research in:
§ Data Integration (Semantic Web, Linked Open Data, etc.)
§ Data Analysis (Data Mining e.g. visitors’ trajectories, Online Social
Networks Analysis, etc.)
Ø LRMH (Research Laboratory for Historical Monuments)
§ Conservation-restoration of monuments
Ø CRCC (Research Center of the Conservation of Collections)
§ Conservation of books, photos, leather objects, etc.
The project was funded for 3 years by the French Heritage Science
Foundation, Excellence Laboratory PATRIMA, and for the 4th year it will be
funded by the French Ministry of Culture.
3.
4.
5. E Provide an unified global access to data related to techniques
and practices in conservation-restoration processes
E Facilitate the knowledge exchange in conservation-restoration
domain, without moving the data held by the conservation-
restoration actors in a more centralized location, nor by merging the
internal schemas of involved data in a uniform common one
7. Ø Conservation-restoration process
v Achievement of scientific studies
(examination, diagnosis, observation,
analysis, etc.);
v Cooperation between different actors
operating in different domains
(conservators/restorers, scientific
researchers, archivists, etc.);
v Clear understanding of the cultural object
and its intrinsic properties (typology,
shape, dimensions, material constitution,
etc.);
v Careful study of the context (location,
environment conditions, conservation
state or degradation, etc.) and history
(origin, phenomena and experienced
events, etc.) of the cultural object;
v An ability to detect interactions that rise
among these elements.
8.
9.
10. Ø Databases Heterogeneity
o Cultural Heritage laboratories with different specificities
o Databases developed with different requirements
o Conservation-restoration data recorded with different formats as well as
different conceptual and structural methods
Ø Lack of semantic power
o Database structures focusing mainly at the syntactical level: relational, semi-
structured (XML annotated documents) or unstructured (texts or images)
Ø Terminology heterogeneity
o Difficult to reach an agreement on a “clear and consistent” terminology
o Difficult to connect databases while avoiding ambiguity, since terms used for
describing the recorded conservation-restoration data may have different
meanings and different lexical values in different databases
11. Building a ontological model intended to capture the existing data
semantics, and to provide a unified understanding of conservation-
restoration data
Anchoring the involved sources to the ontological model
Building a query-answering process allowing to get
information from the different conservation-restoration semantic
databases, through the ontological model
13. PARCOURS ontological model is
composed of a triple
Op = ⟨ (Ot, Oc ), Ter ⟩ where:
– The couple (Ot, Oc) is composed by,
respectively, a top-level ontology Ot
(CIDOC-CRM), and a core ontology Oc
(PARCOURS core). PARCOURS core
ontology is connected to the CRM_sci
for representing at a very fine level of
granularity the specificities of scientific
observations.
– Ter consists of a set of domain-
specific thesauri (controlled
vocabulary terms) intended to provide
an unambiguous conservation-
restoration terminology.
PARCOURS Ontological Model
14. Ø Thank you, Achille, for so well presenting CIDOC-
CRM yesterday! J
Ø CRM_sci (“Scien&fic Observa)on Model”) integrates
metadata about scien)fic observa)on, measurements and
processed data in descrip)ve and empirical sciences
15. Ø Snapshot from the ontological model
Concept from CIDOC-
CRM
Concept from CRM
SCIENCE
Concept from
PARCOURS
16. Ø Anchoring the sources to the global ontology
Ø Each participating source is autonomous and stores its data
without changing its internal formats
Ø Each participating source must build a conceptual representation
of its data, at least for the part that should contribute to the
integration process
Ø Each source is free to choose the method to bring its data to its
local semantic database
– Whether possible by using suitable existing tools: e.g. Ontop [Bagosi et al.
2014]
– Otherwise by implementing its own ad hoc algorithm according to its
database format
– Regardless the solution chosen by a source, the generated repository must
be updated and conceptually sound with respect to the PARCOURS global
ontology
17. Different Actors and/or
applications
R2ML
Mappings A
P
I
Json
ad hoc
algorithm
ONTOP ETL
Query Processing Engine
Sparql EndPointSparql EndPoint
Relationnal
Database
Sesame
RDF
Repository
….
Source1 Sourcen
PARCOURS global
Ontology
Request
RDF
Repository
Response