=Paper=
{{Paper
|id=Vol-2949/paper1
|storemode=property
|title=An Approach to Semantic Representation and Modeling in the Development of De rerum natura Digital Exhibition
|pdfUrl=https://ceur-ws.org/Vol-2949/paper1.pdf
|volume=Vol-2949
|authors=Carlo Teo Pedretti
|dblpUrl=https://dblp.org/rec/conf/swodch/Pedretti21
}}
==An Approach to Semantic Representation and Modeling in the Development of De rerum natura Digital Exhibition==
https://ceur-ws.org/Vol-2949/paper1.pdf
An Approach to Semantic Representation and
Modeling in the Development of De rerum
natura Digital Exhibition
Carlo Teo Pedretti1
Department of Classic Philology and Italian Studies, University of Bologna, Via
Zamboni 33, 40126 Bologna, Italy
carloteo.pedretti@studio.unibo.it
Abstract. The purpose of this paper is to present the implementation
of De rerum natura Exhibition (DRN)1 , a digital exhibition, with par-
ticular regards to Semantic Web technologies and design patterns which
allow the seamless representation of heterogeneous contents and the cre-
ation of dynamic museal paths. In particular, this study presents the
implementation solutions that have emerged during the development of
a digital application concerning Vedere l’Invisibile 2 , an exhibition which
follows Lucretius’ De rerum natura and its reception in contemporary
art. The peculiarity of the exhibition is the presence of heterogeneous
data sources such as tangible or intangible cultural items (e.g. paint-
ings, performances) and textual documents (e.g. De rerum natura and
its translations). Starting from the aforementioned sources, the work ex-
plores modern technological solutions in order to deliver a system that
can be both flexible in terms of requirements, reusable and ephemeral.
Results show that by adopting a wrapper design pattern, a functional
implementation of the system described above can be delivered without
affecting original sources and successfully integrating different contents
such as textual documents and cultural items through semantic tech-
nologies.
Keywords: digital exhibition · heterogeneous collections · Open An-
notation · Semantic Web
1 Definition and Scope
The project revolves around the digitalization of Vedere l’Invisibile, an exhibi-
tion held in Bologna in 2017 with the shared efforts of scholars from different
domains. The aim of the exhibition was to celebrate six centuries from the dis-
covery of Lucretius’ De rerum natura and specifically the influence it had on
contemporary arts. In particular, De rerum natura exposes the principles of Lu-
cretius’ philosophy of nature, recurring to a number of images and metaphors
1
https://github.com/friendlynihilist/lucretius
2
https://eventi.unibo.it/riscoperta-lucrezio
Copyright © 2021 for this paper by its authors. Use permitted under Creative Commons
License Attribution 4.0 International (CC BY 4.0).
2 Carlo Teo Pedretti
that inspired crucial innovations in visual culture. While the marks left by the
poem in Renaissance and modern art are known, its influence on contemporary
art is somewhat left behind by scholars. Nevertheless, themes of De rerum natura
– the caducity of nature, the frailty of knowledge and the disorientation of man
inside an indifferent cosmos – have become central in our time and therefore it is
not surprising that several original voices have tried to express them by means
of different media. Contemporary vision of De rerum natura is declined through
several media such as drawings, paintings, engravings, sculptures, films, music
and performances. Moreover, the exhibition contains works and installations by
several famous contemporary artists (Baj, Paolini, Mezzaqui etc.) and some of
the most important Italian poetic translations (Sanguineti, Orelli, De Angelis)
of De rerum natura. The exhibition is the result of an intense research activity
that has involved philologists, art, science and books historians, offering the op-
portunity of dealing with an interdisciplinary project around an enthralling new
theme.
DRN starts from these heterogeneous collections with the aim of creating an
interactive digital exhibition using a modular framework and a light implemen-
tation of Semantic Web technologies. Differently from other exhibitions, Vedere
l’Invisibile has been built around a specific literary work (Lucretius’ De rerum
natura) and its propagations in other domains. Therefore, De rerum natura influ-
ence upon other cultural heritage domains also maintains a primary importance
in the design phase of the digital exhibition experience. Not only the final users
should be able to discover works of art connected to or influenced by De rerum
natura, but also the narrow bond between textual and artistic expressions should
be described in an innovative and engaging way.
Several technologies have been employed in the development of the digital
exhibition and in particular Omeka S3 and TEI Publisher4 have been employed
for managing respectively digitalized tangible or intangible cultural items and
textual documents. Furthermore, RDF(S)5 , Web Annotation Data Model6 and
CIDOC CRM7 have been employed in the creation of a data model for describing
multi-layered, narrative paths designed by the curator.
Within this context, this study primarily aims at answering the following
research questions:
– Is it possible to build a narration which involves heterogeneous data sources
within the context of a digital exhibition without recurring to a vertical, silo
approach?
– To what extent can an adapter design pattern be capable to process and
represent information homogeneously, harmonizing contents and models?
Results show that by adopting a wrapper design pattern, a functional im-
plementation of the system described above can be delivered without affecting
3
https://omeka.org/s/
4
https://teipublisher.com/index.html
5
https://www.w3.org/TR/rdf-schema/
6
https://www.w3.org/TR/annotation-model/
7
http://www.cidoc-crm.org/
Semantic representation and modeling in DRN 3
original sources and successfully integrating different contents such as textual
documents and cultural items through semantic technologies.
2 Background and Significance
Temporary exhibitions are a traditional and successful cultural activity that
art museums have sponsored for decades. As their name implies, the life-cycle
of these events is ephemeral and short-lived. Nonetheless, the development of
digital, online versions of exhibitions and their availability in time with extended
features that cannot be replicated in a physical environment means that the
impact of the adjective ‘temporary’ has started to fade [3].
Furthermore, the implementation of technologies and standards borrowed
from other cultural domains, such as metadata and ontologies, signals the rise of
a new era in terms of communication and dissemination of museal contents. In-
deed, while Semantic Web technologies are used to connect and describe all sorts
of data for business purposes, the vast cultural heritage domain in its hetero-
geneity has profited from implementing such technologies for archiving, curating
and disseminating its objects, opening up to unprecedented possibilities [4].
In particular, apart from being an answer to the need of preserving the ma-
terial counterpart, Semantic Web and Linked Open Data (LOD) have been used
successfully in many domains to foster exploration and discovery of cultural her-
itage objects [2]. LOD brings the possibility of linking datasets from different
cultural institutions, providing new horizons of interconnection and accessibility
to the public. These practices have contributed to build a shared knowledge base
that allows reusability through different applications and domains. Many ontolo-
gies and Knowledge Organization Systems (KOS) have been developed in order
to exploit that knowledge base. On the other hand, the ontological heterogeneity
of cultural heritage entities requires different approaches in terms of knowledge
representation and data modeling [1].
Finally, as narrative can be identified as the distinctive feature of exhibi-
tions with respect to other cultural institutions, the technologies described above
should also serve the purpose of designing environments in which narrations and
curatorial intents merge seamlessly into a constant dialogue between the items
and the visitors of a collection.
3 Problems
Indeed, one obstacle in the development of digital exhibitions in general lies in
the modeling and representation of the items that form the exhibition itself. In
particular, DRN draws upon two radically different data sets: items (paintings,
engravings etc.) and textual documents (transcriptions and translations).
Defining the relationships between items and literary works is not always
linear and often involves hermeneutical efforts and precise choices in terms of
granularity and serialization of data. In these cases, recurring to data silos or
4 Carlo Teo Pedretti
vertical approaches in terms of information management inevitably leads to a
lack of reusability and flexibility of the system.
Another challenge is represented by the nature of the items themselves. For
instance, some items could be part of a larger set or belong to the same series
(as in the case of Baj’s engravings), while others are different versions of the
same work (as in the case of Paolini’s Casa di Lucrezio). For these reasons,
modeling the exhibition should take into account meronymy, sequentiality and
temporality.
Finally, an exhibition should avoid when possible passive fruition and there-
fore has to introduce mechanisms to enhance active discovery and browsing, both
through advanced filtering, multiple views and dynamic paths.
These prerequisites have been taken into account in the preliminary phases of
development of DRN by sketching both multiple possible views on the collections,
and by designing a storytelling component that offers dynamic museal paths
based on semantic relationship between items and collections.
4 Method
DRN aims to overcome lack of reusability and reproducibility by building a
flexible framework that acts as a wrapper on top of sources. Moreover, it imple-
ments Semantic Web technologies consistently in order to manage heterogeneous
sources without recurring to vertical approaches that will in return hinder its
true potential. By acting as a wrapper, DRN aims to create a homogeneous space
of fruition for the user, where information organization and management with
regards to exhibition contents becomes independent from the actual sources.
Having taken into consideration the nature of the exhibition and its collec-
tions, DRN has been designed and implemented using a two-tier architecture,
which is based on the separation of the data logic and the presentation logic,
as shown in Figure 1 in the following page. In simple software architecture, the
presentation layer or user interface layer runs on the client side while the dataset
layer gets executed and stored on the server side. As such, there is no business
logic layer or immediate layer in between client and server; in fact, the applica-
tion logic is built in parallel with the presentation layer. The frontend framework
manages both the presentation layer as well as the application logic, operating
as an “ephemeral” wrapper on top of data sources, which it consumes for cre-
ating dynamic views. An adapter pattern design has been chosen primarily for
lightness and flexibility.
Specifically, data is stored in two different repositories based on its format.
Textual documents and fragments are XML-TEI encoded and stored inside an
instance of TEI Publisher, an eXist-db application, that manages and publish
TEI documents in HTML and other formats through an ODD file, based on
the TEI Processing Model. Multimedia digital objects are mostly Dublin Core
encoded and managed by Omeka S. These data sets are then exposed through a
REST API as JSON-LD multi-dimensional arrays. Moreover, RDF models are
stored in a quad store subsystem and dynamically parsed and extracted.
Semantic representation and modeling in DRN 5
Fig. 1. DRN digital exhibition architecture model
6 Carlo Teo Pedretti
The application logic is entirely made through Angular, and clustered in com-
ponents such as helpers and parsers. Therefore, Angular manages both the appli-
cation logic and the presentation layer. Helpers and parsers work as adapters for
retrieving and transforming the various data formats (JSON-LD, RDF, XML) in
order to be consumed and dynamically presented through the DOM. As stated
before, one of the main goals of this digital exhibition is to create sophisticated
museal paths based on semantic relationships between textual fragments and
items of the collections. In order to do so, further technologies have been intro-
duced to design and manage semantic relationships.
First of all, it is not always possible to embed metadata or annotations into
documents as in the case of the literary works that compose this digital exhi-
bition. In these scenarios, standoff formalisms support the separation between
the document and its annotations. The Web Annotation Data Model has been
developed by the Open Annotation W3C Community Group and provides a
RDF-based approach for standoff annotation of web documents, serialized in
JSON-LD. While Web Annotation has been designed to be used across different
formats, the most common scenario is to connect text fragments to a single web
resource.
The Web Annotation provides reified representation of annotated elements
and their relative annotations: Body, Target and Annotation are discrete re-
sources and can be further specified with more properties and relationships.
oa:hasSource and oa:hasSelector can respectively define the resource from
which the annotated element is refined from and the object of the relationship
that describes the segment or region of interest within the source resource. An-
other crucial selector is the oa:XPathSelector, that uses XPath for identifying
elements of an XML document (e.g. a DOM tree). In the case of the XML-TEI
encoded documents used in DRN, the aim is to create a relationship between an
item of the exhibition and a section of the XML-TEI document and successively
between the XML-TEI document and the respective DOM counterpart.
Secondly, CIDOC CRM, as well as RDF(S), have been integrated in order
to develop complex museal paths. In particular, as shown in Figure 2 in the fol-
lowing page, a given resource (”E73-Information Object”) refers to (”P67-refers
to”) a Target that describes an item of the collection (o:Item) or an annotation
(oa:Annotation). A given resource always has a type (”P2-has type”), a time-
span (”P4-has time-span”) and note(s) attached to it (”P3-has note”). Further-
more, a list or ordered collection of resources defines a topic (”E89-Propositional
Object”) that can contain named entities (rdfs:seeAlso) and interpretative
note(s) attached to it (”P3-has note”) that in turn can point to internal or
external resources depending on the context. In this context, a topic offers in-
formation on a particular subject. For example, as shown later in this study, the
topic named Le acqueforti di Baj presents curatorial notes regarding a series of
engravings made by Enrico Baj. In turn, a collection of topics defines a story or
(museal) path (”E89-Propositional Object”).
As an example, La peste d’Atene is a path (a story) designed with the aid
of curators and domain experts, presented as a succession of topics (enriched
Semantic representation and modeling in DRN 7
Fig. 2. UML diagram representing the data model
with interpretative notes). La peste d’Atene refers to the final part of De rerum
natura. In addition to being philosophically fundamental within the Lucretian
speculative system, the ending of De rerum natura bears dramatic suggestions
which were inspiring for artists of the exhibition such as Enrico Baj and Teresa
Procaccini, among the others.
By describing the model as a series of triples using CIDOC CRM and RDF(S),
it is possible for example to refer a path to a collection of topics, using an RDF
List. In particular, La peste d’Atene presents nine different topics:
@prefix ecrm: .
@prefix rdfs: .
@prefix rdf: .
a ;
ecrm:P102_has_title .
a ecrm:E35_Title ;
rdfs:label "La␣peste␣d’Atene" .
ecrm:P67_refers_to ;
rdf:_1 ;
rdf:_2 ;
[...]
rdf:_9 .
8 Carlo Teo Pedretti
A topic is defined by a title and a body. It could also contain references to Named
Entities and interpretative notes:
@prefix ecrm: .
@prefix ex: .
@prefix rdfs: .
@prefix rdf: .
a ;
ecrm:P102_has_title ;
ex:hasNamedEntity ;
ecrm:P3_has_note .
a ecrm:E35_Title ;
rdfs:label "Le␣acqueforti␣di␣Baj" ;
rdf:value "Le␣acqueforti␣di␣Baj" .
rdfs:seeAlso ;
a ;
rdfs:label "Arturo␣Schwarz" .
a ;
rdf:value "Nell’aprile␣del␣1958,␣Arturo␣Schwarz␣pubblica␣De␣
rerum␣natura,␣una␣cartella,␣stampata␣in␣51␣esemplari,␣che␣
raccoglie␣36␣acqueforti␣di␣Enrico␣Baj␣[...]" .
Moreover, and most importantly, target items could also have references to tex-
tual parts of De rerum natura, as in this example:
@prefix rdfs: .
@prefix oa: .
@prefix rdf: .
rdfs:seeAlso .
a oa:Annotation ;
oa:hasBody ;
oa:hasTarget .
oa:hasSource ;
oa:hasSelector .
Semantic representation and modeling in DRN 9
a oa:XPathSelector ;
rdf:value "//*/*[position()>63␣and␣position()<79]/text()" .
In particular, OA annotations allow not only to annotate textual parts and
associate them with items, but also to set up the core which made possible
the realization of thematic paths. In the design pattern, annotations assume the
dignity of an autonomous resource: in fact they are the contents of the exhibition.
As demonstrated above, the data model has been designed with a selection
of well-known standards (CIDOC CRM and RDFS) in mind, and this allows to
obtain a simple and light framework for describing the resources semantically.
In addition, by not implementing a custom ontology, reusability and interoper-
ability of the framework are pursued.
Fig. 3. An example with actual data from the digital exhibition
Figure 3 shows an implementation with real data from the exhibition. In the
example, a collection of resources that refers to items and textual parts (e.g.
Donna Urlante and other engravings from the series Storie della morte by En-
rico Baj) defines the topic titled Le acqueforti di Baj, which is here presented
accompanied by a critical text by the curator and a link to a Named Entity (Ar-
turo Schwarz, the publisher of Baj’s series of engravings about). Le acqueforti di
Baj, along with other topics such as La peste d’Atene secondo Teresa Procaccini,
defines a collection, which in turn determines La peste d’Atene as a path. Thus,
10 Carlo Teo Pedretti
a path acts as an aggregator or container of resources and semantic relationships
between data.
5 Results and Further Developments
The main goal of DRN digital exhibition was to offer customized museal paths
based on semantic relationships between textual parts and items to the final
users.
One of the limits that has been encountered in the generalization phase
concerns the alignment between XML-TEI sources and the HTML presentation
structure, due to the constraints imposed by the integration of TEI Publisher. In
particular, a solution had to be implemented that would allow the TEI sources to
integrate with the presentation layer without losing information on their struc-
ture. The answer was found using Web Annotation Model. Nonetheless, through
the implementation process of the prototype, it has been shown that a wrapper
approach (opposed to a vertical, silo-like approach) is not only achievable in this
context, but also useful in order to build a narrative which involves different
digital sources. In particular, the adapter design pattern has proved to be an
optimal solution in order to represent information and harmonize heterogeneous
contents with their models while leaving the sources untouched. In this way,
reusability and interoperability have been addressed.
Fig. 4. An overview of the items collection from DRN digital exhibition
While this is not the focus of this work, a very brief overview of the pre-
sentation layer could prove useful in order to present a concrete application in
the cultural heritage domain of the architecture presented above. The design of
interfaces and browsing aids for digital collections is a crucial step for delivering
Semantic representation and modeling in DRN 11
a successful user experience, even more so if considering that discovery is one
of the most important elements in applications of this sort and ”the interface
[...] often a user’s only point of contact with the collection” [5]. DRN interface
has been designed following primarily Whitelaw’s ”generous interfaces” ethos as
a principle capable of providing a critical perspective on the collection access
point.
DRN is first and foremost a digital exhibition, thus the figurative element
plays a central role in the presentation of its contents. Figure 4 shows the thumb-
nail preview of every item of the collection, which presents an overview on the
individual elements, enabling identification at a glance. A filtering and facets
system allows to refine the results employing item metadata. By clicking on a
thumbnail, the user can enter the record page which contains media, metadata
and, eventually, the verses significantly related to the item. Furthermore, by
clicking on the verses, user is redirected to the TEI Publisher interface, in which
corresponding verses are highlighted; also, a popover with the thumbnail and
basic metadata of the related item appears.
Fig. 5. Story component view from DRN digital exhibition
Finally, Figure 5 shows the story component view. Stories are accessible
through a dedicated voice in the navigation bar. After choosing a path (e.g.
”The plague of Athens”), the user can navigate it both from a temporal point
of view, by using the timeline, or by selecting the topics (e.g. ”Le acqueforti di
Baj” etc.) which act like chapters. Every topic is also connected to particular
items of the collection and offers both internal (e.g. resources, textual parts) and
external references (e.g. Named Entities connected to Authority Files).
Nonetheless, as this remains a prototype, further testing needs to be carried
out in order to evaluate the effectiveness of the system in communicative terms.
12 Carlo Teo Pedretti
In particular, it should be evaluated if this design pattern allows the visitors
of the digital exhibition to reach a more thorough understanding both of the
collection and its context.
From a technological point of view, added views can be implemented as
further developments. In particular, a force-directed graph could be an optimal
solution in terms of data visualization for representing relationships between
institutional actors such as artists and curators by employing metadata and
Semantic Web technologies.
To conclude, DRN digital exhibition proves that with flexible choices in terms
of design implementation, an adaptive, reusable, ephemeral and disposable so-
lution can be achieved, that allows heterogeneous contents and models to be
described and represented harmoniously.
References
1. Cimiano, P., Chiarcos, C., McCrae, J.P., Gracia, J.: Linked open data for cultural
heritage. Representation, Generation and Applications, pp-229–262. Springer Inter-
national Publishing, Cham (2020). https://doi.org/10.1007/978-3-030-30225-2
2. Marden, J., Li-Madeo, C., Whysel, N., Edelstein, J.: Linked open data for cultural
heritage. In: Proceedings of the 31st ACM International Conference on Design of
Communication, pp-107–112. (2013). https://doi.org/10.1145/2507065.2507103
3. Mateos-Rusillo, S.M., Gifreus-Castells, A.: Museums and Online Exhibitions: A
Model for Analysing and Charting Existing Types. Museum Management and Cu-
ratorship 32(1), 40–49 (2017)
4. Sahle, P.: What is a Scholarly Digital Edition?. In: Digital Scholarly Editing: The-
ories and Practices, pp. 19–39. Open Book Publishers, Cambridge (2016)
5. Whitelaw, M.: Towards Generous Interfaces for Archival Collections. Comma 2,
123–132 (2012). https://doi.org/10.3828/comma.2012.2.13.