=Paper=
{{Paper
|id=Vol-1509/submission_5
|storemode=property
|title=Using Ontologies as a Faceted Browsing for Heterogeneous Cultural Heritage Collections
|pdfUrl=https://ceur-ws.org/Vol-1509/ITALIA2015_paper_5.pdf
|volume=Vol-1509
|dblpUrl=https://dblp.org/rec/conf/aiia/TomasiCDL15
}}
==Using Ontologies as a Faceted Browsing for Heterogeneous Cultural Heritage Collections==
https://ceur-ws.org/Vol-1509/ITALIA2015_paper_5.pdf
Using ontologies as a faceted browsing for heterogeneous
cultural heritage collections
Francesca Tomasi1, Fabio Ciotti2, Marilena Daquino3, and Maurizio Lana4
1
Department of Classical Philology and Italian Studies, University of Bologna, Italy
francesca.tomasi@unibo.it
2
Department of Humanities, University of Roma Tor Vergata, Italy
fabio.ciotti@uniroma2.it
3
Multimedia Research Resource Centre (CRR-MM), University of Bologna, Italy
marilena.daquino2@unibo.it
4
Department of Humanities, University of Piemonte Orientale, Italy
maurizio.lana@uniupo.it
Abstract. In this paper we present a project regarding the possible use of multi-
ple and interconnected OWL ontologies (GO!, HiCo, and Proles) in order to
explore the semantic content of heterogeneous digital collections (a digital li-
brary, a full-text scholarly edition, and a relational database) in the cultural her-
itage domain (Geolat, Vespasiano da Bisticci Letters, and Zeri photo archive).
The aim is to discover knowledge by revealing, through facets, possible latent
connections – or even contradictory statements – between data, moving from
person, places and dates in an event-centric dimension determined by a context-
oriented perspective.
Keywords: ontology matching, data modelling, facets, context, cultural herit-
age.
1 Introduction
Ontologies are an important tool for conceptualizing knowledge in a domain-oriented
perspective, as in the cultural heritage scenery. But, in general, ontologies could be
conceived and developed in order to potentially achieve distinct kind of goals, e.g.: to
annotate full-texts according to a semantic model [1, 2]; to formalize concepts and
constraints in a domain (e.g. digital editions) by using a data-centric approach [3, 4];
to export relational databases in open linked datasets, by converting tables and fields
in classes and predicates [5]. Similarly, as known, collections in cultural heritage
domain collect data from different areas in humanities (e.g. literature, art, history,
etc.), were ontologies likewise could cover different roles and convey multiple func-
tions.
As described later, our research groups developed, in these last few years, many
different ontologies, and realized various heterogeneous collections in cultural herit-
age domain (see references section for relevant papers). It was then a challenging
objective to think about a possible interconnection between these ontologies, having
the digital collections as a suite of complex objects, i.e. a knowledge base, to test the
ontologies.
Starting from these considerations, the attempt here described is to reflect on pos-
sibilities offered by ontologies as cognitive tools. In fact, efforts in modelling could
also be exploited for defining ‘exploratory’ methods. Collections can be browsed by
using classes as conceptual categories, i.e. by following a taxonomical approach.
Predicates define dynamic filters able to express multi-level relationships. Hierar-
chical, associative, and equivalence relationships between instances and therefore
classes are able to define ‘facets’.
In order to test this theoretical model, this feasibility study aims at creating a dia-
logue between different digital collections and to use ontologies, developed ad hoc for
single collections, as a tool for browsing all the heterogeneous complex objects. Cul-
tural objects are keys for creating relationships both at the conceptual and the physical
level: original objects, digital objects, subjects of the objects, and interpretations of
the objects.
The semantic environment we imagine focuses on specific potentially connected
classes and predicates, which could be defined as the facets for browsing the whole
collections. Facets help in searching for identities or affinities - but also potentially
dissimilarities - between the involved complex objects. In particular, the scenario we
imagine aims at defining facets starting from the most common categories in the cul-
tural heritage domain:
People. Each person has a specific role in context; i.e. the same person could cover
different roles, depending on the context (e.g. the document) where the entity acts.
People are agents: they reply to the question ‘who’;
Dates. Dates give consistency to actions that involve people in a real or ‘virtual’
place (document/record/fragment). Dates define temporal entities: they reply to the
question ‘when’;
Places. Places are useful to identify real or virtual spaces in which events happen.
Places are identifiable spaces: they reply to the question ‘where’.
Relationships between classes are defined by using an event-oriented approach:
people act in a specific date and place, creating the event. Actions reply to the
question: ‘what’. The event is related to the context, i.e. the context determines the
connection between classes.
Each action aiming at recognizing people, dates, places and event in a context is an
interpretative action, made by an agent (e.g. the editor) through his/her knowledge,
background or by consulting other sources. The interpretative process replies to the
question ‘why’.
Classes representing cultural objects have to be managed according to FRBR1 model,
following the idea of a hermeneutical approach based on a multiple level analysis.
People, dates, places and events have all to be described as referred to the specific
level of the cultural heritage object: work, expression, manifestation, item.
1
Functional Requirements for Bibliographic Records (FRBR),
http://www.ifla.org/publications/functional-requirements-for-bibliographic-records
Starting from these assertions we choose, in detail, to put in dialogue three ontolo-
gies2, potentially all related to the formalization of a wide and complex field in the
cultural heritage domain (described in section 2):
GO [1, 2, 6], defines space and places in a geographical dimension;
Political Roles (PRoles) Ontology [3, 4], defines people with a role in a
space/time-indexed situation;
HiCO ontology [3, 7], manages relationships between cultural objects and any
entity considered an interpretation about the object itself.
These three ontologies have to be experimented on three chosen collections3 that rep-
resent three different, but related, projects, both for content and domain (described in
section 3):
the digital library: Geolat4 [1, 2, 6].
the digital edition: Vespasiano da Bisticci’s Letters5 [8].
the relational database: Zeri Photo Archive6 [5, 9].
In order to complete this project some operations will have to be further defined (de-
scribed in section 4). We will also have to define a possible framework to explore
functionalities of our method, e.g. a semantic repository for semi-structured data, and
an environment for the faceted browsing [see e.g. 10] that allows classes and predi-
cates to be transformed in facets.
2 Ontologies for cultural heritage modelling
Ontologies here chosen represent different parallel activities of our research group in
digital humanities field. The domain is in fact the humanities; in particular the
knowledge base on which we modelled concepts is represented by literary texts and
documents, but also descriptive entries for cultural objects. The aim of the project is
to reflect on the most common features of the analysis in a literary dimension: geo-
graphical and historical information (section 2.1); levels and methods of interpretation
used by an editor of resources in the field of humanities (section 2.2); roles of people
involved in actions described in the sources (section 2.3).
2
In particular, as referred in the final references, GO! is by M. Lana and F. Ciotti; HiCO is by
M. Daquino, S. Peroni, and F. Tomasi; PRoles is by M. Daquino, S. Peroni, F. Tomasi, and
F. Vitali.
3
In detail: Geolat is by M. Lana, D. Magro, F. Ciotti, with C. Meini, M. Benzi, and G. Vanotti,
is funded by Fondazione Compagnia di San Paolo; Vespasiano da Bisticci’s Letters is by F.
Tomasi; Zeri Photo Archive is by C. Gognano, F. Mambelli, S. Peroni, F. Tomasi, and F.
Vitali.
4
Geolat, http://www.geolat.it
5
Vespasiano da Bisticci, Letters, http://vespasianodabisticciletters.unibo.it
6
Fondazione Zeri, Photographic Archive, http://www.fondazionezeri.unibo.it/catalogo
Each single ontology has already imported classes from other similar domains,
and proceeded with the mapping on the most common ontologies in the cultural herit-
age domain (especially EDM7, CIDOC-CRM8, DC9 and DCMI terms10).
Such models, if gathered, can represent a complete and well formalized point of
view on approaches and methodologies used by humanists when dealing with several
and heterogeneous cultural objects (i.e. texts and digital collections, works of art and
related metadata). These ontologies are able to represent a huge part of the formal
activities acted by editors considering their hermeneutical approach. Shared entities
among ontologies are people, places, dates and events – i.e. all significant and mini-
mal requirements for describing a complex scenario – and each of them further exam-
ine a specific issue, fundamental for an overall awareness of the domain of cultural
objects description.
As we said, ontologies in general – and the chosen ontologies in particular – were
developed for different use and for documenting different contexts. The project wants
to explore connections between these ontologies, in order to use them as a mean for
browsing information. The aim is to test these ontologies on the above-mentioned
collections and to highlight possible re-usability of a single facet in a different context
(i.e. another dataset, created with different purposes and a different data structure).
The alignment will be necessary in order to guarantee the semantic portability of the
resulting model (section 4), even at the moment the ontologies represent individual
categories for the exploration.
2.1 GO! Geolat Ontology
GO! is a Geographical Ontology that is now developed inside the Geolat project (see
section 3.1). This last has the aim to annotate a digital library of latin texts digilibLT11
where annotation is realized using GO!. This ontology is built ad hoc for Geolat, reus-
ing data offered by Pleiades gazetteer12, and establishing relationships with other rele-
vant geographical ontologies (first of all geoSPARQ 13 and GeoNames 14 ). GO! is
structured as a two-tier model: a T-box modelling geospatial classes of locations, their
properties and their relationships and an A-box with geospatial information about
individual places and location. Four modules are defined in GO: GO-TOP: general
module for top-level concepts; GO-FAR: For Ancient Resources; GO-PHY: Physic
geography (e.g. mountains, rivers); GO-HUM: geosocial module describing Human
Artefacts (e.g. cities) and social structures (e.g. formally defined regions, territories)
7
Europeana Data Model (EDM). Documentation, http://pro.europeana.eu/page/edm-
documentation
8
CIDOC Conceptual Reference Model (CRM), http://www.cidoc-crm.org/
9
Dublin Core (DC), http://dublincore.org/
10
Dublin Core Metadata Initiative (DCMI) Terms, http://dublincore.org/documents/dcmi-
terms/
11
digilibLT, http://www.digiliblt.uniupo.it
12
Pleiades, http://pleiades.stoa.org/
13
geoSPARQL, http://geosparql.org/
14
GeoNames, http://www.geonames.org/
2.2 PRoles. Political Roles Ontology
We have developed the PRoles Ontology15 to explore new possibilities in the repre-
sentation of authority records. The aim was to formalise complex relationships, such
as agents’ political roles and events in which these agents are involved, with a specific
role, as attested in full-text sources. Initial targets were indeed related to sources deal-
ing with historical events and relevant related politicians, in order to describe useful
information in archival and historical context. At the same time this approach was
adopted in order to define a deeper analysis of relations among people (corporate
bodies, families and persons as in EAC-CPF16 definition) and documents where they
are cited. A particular attention was then devoted to the provenance principle, in order
to formally define the paternity for the assertions.
Except specific features related to this restricted domain of interest, PRoles im-
ports and extends three other models – PRO ontology17, N-ary Participation pattern18
and PROV-O19 – which allow to describe a wide range of information extracted from
full-text sources stakeholders can be interested to. In particular with PRoles it could
be possible to attribute a role to an agent, to specify the event in which the agent is
involved, to define the responsibility of the attribution (fig. 2).
Although PRO ontology had been thought in principle for an application in the
publishing domain, it has been developed so as to accommodate any kind of role,
regardless the domain of interest. In particular, PRO defines a class to specify roles an
agent can holds, pro:Role, and a class for representing role attributions as reified rela-
tionships, i.e. individuals of the class pro:RoleInTime, which allows to describe
agents’ having a role in a precise interval and within a particular context (such as in
some organisation or place, on a document or with respect to other agents).
Although PRO provides a first formalization of relationships (describing someone
holding a role within a particular context), the description of information that can be
extracted from full-text often needs an additional level of contextualisation, like de-
scribing agents participating to events (located in time and in space) with a particular
role. In order to enable such descriptions, another model was reused to include agents
with relationships in events, i.e., the N-ary Participation ontological pattern. This
pattern describes, mainly, individuals of the class nary:NaryParticipation, which
allows modelling any object as a participant in an event, i.e., an agent who partici-
pates for a specific period of time in an event, holding a time-indexed political role
and relating with other objects (agents, places, sources, etc.).
15
Political Roles Ontology (Proles), http://www.essepuntato.it/2013/10/politicalroles
16
Encoded Archival Context - Corporate Bodies, Persons, and Families (EAC-CPF),
http://eac.staatsbibliothek-berlin.de/
17
Publishing Roles Ontology (PRO), http://purl.org/spar/pro
18
Nary Participation, http://ontologydesignpatterns.org/wiki/Submissions:Nary_Participation
19
PROV-O: The PROV Ontology, http://www.w3.org/TR/prov-o/
Fig. 2 Proles model: role attribution, participation to events, provenance
2.3 HiCO. Historical Context Ontology
HiCO20, is an OWL 2 DL ontology created in order to define a formalized, shared and
exchangeable model for describing the context of cultural heritage objects and the
workflow for stating authoritative assertions about such information. When formaliz-
ing any sort of assertions that can be questionable, we are stating something about an
object, i.e. extracting something from its content. Each assertion is a subjective au-
thors’ ‘reading’ which involves a specific layer of the source wherefrom the reading
belongs to, e.g. the text, or better the expression of the source of interest. This com-
pels a multilayers representation of the text – or of the cultural object of interest –,
necessary for clearly defining an interpretative process as meta-contextual level for
provenance of assertion. It’s important to have in mind that a text is a multi-faceted
object that has to be treated at different levels of analysis. HiCo ontology, according
to FRBR model, attempts to solve part of this issue, and considers other and more
specific questions related to the interpretative process. So, HiCo starts from the first
formalization of provenance statements in PRoles (2.2) and extends the analysis of
required entities involved in the interpretative process, defining a more detailed work-
flow.
In particular HiCO – that reuses FRBR, Pro, N-ary and Prov-o – allows to de-
scribe the agents’ interpretative act by specifying ‘types’ and ‘criteria’ of the interpre-
tation and, with the help of the CiTO ontology21, to create relationships between in-
terpretations referred to the same cultural heritage object, made by different agents
(e.g. editors).
20
HiCo, Historical Context Ontology, http://purl.org/emmedi/hico
21
CiTO, the Citation Typing Ontology, http://purl.org/spar/cito/
The general HiCO workflow could be synthesized in: an agent (3) extract as RDF
statements (7) some contents form a source. Such contents are born out from an inter-
pretative activity (1), of a certain type (6), made by another agent (e.g. an editor) by
using a particular criterion. The agent (e.g. editor or cataloguer) makes interpretations
(1) in a new work (5) regarding the cultural object (4). The interpretation act declares
the authority on which it’s based and/or the agreement with another interpretation act
(fig. 3).
Fig. 3 HiCO model
3 Cultural heritage collections: the knowledge base
Resources on which the project is working represent three different complex object
collections, or even three different models: a digital library, a digital edition, a rela-
tional database. All these collections have a technological common base: they are
available as XML datasets, and the information regarding people, places, dates and
events have been identified with URIs. The idea is to further analyse and collect such
resources, in order to study how objects can be browsed through the use of the aligned
ontologies, exploring possible new information generated by alignment itself.
These issues are addressed meanwhile datasets are refined for their publication. In
this phase, rethinking in a wide perspective their initial conceptualization and formal-
ization (i.e. the adoption of an ad hoc ontology to model information), will enable
more information discoverability whatever will be the implementation. The choice of
gathering below described projects is justified by the wide field of research they aim
to describe, as together they cover a heterogeneous range of information which could
be of interest for other research groups and feasible in other use cases.
As we said above, with respect to the choice of the ontologies, also the collections
presented here are the result of our research group activities. These collections cover
multiple aspects of humanistic interest: literary texts (section 3.1), manuscripts and
archival documents (section 3.2), and finally photographs depicting works of art (sec-
tion 3.3). These data exemplify suitable layers of analysis in a domain-oriented per-
spective, representing different typologies of humanistic data. At the same time this
choice is able to describe three different ways of knowledge transmission in a digital
environment: a library, an edition and a catalogue. In order to test the ontologies by
creating a common semantic model starting from people, dates, places and events, all
the collections should reply to the questions: who, where, when, what and why?
3.1 Geolat
The aim of Geolat project is to make accessible the Latin literature through a query
interface of geographic / cartographic type. The work starts from a digital library that
when completed will contain works of Latin literature from its origins to the end of
the Roman Empire (conventional date, the 476 d. C.). This stage involves the integra-
tion of various already existing repository of Latin texts of high philological quality,
which will be integrated starting from their already existing TEI/XML encoding. In a
second phase the works so collected are analyzed at morphological level by means of
a parser (Lemlat of ILC in Pisa) so as to associate with each word its analysis / mor-
phological description, including a first-level identification of proper names done with
NER (Name Entity Recognition). A third level of modelling will be tied to the logical
relationship between textual references (and their annotations by an encoder) and their
referent in the GO! ontology (section 2.1).
3.2 Vespasiano da Bisticci’s Letters
A digital annotated (XML/TEI) collection of letters from the XV century,
sent/received to/by the florentine copyist Vespasiano da Bisticci (fig. 4). The collec-
tion is available in a web environment that focuses on: people mentioned in the doc-
uments; classical latin and greek manuscripts requested/copied/proposed to/by Ves-
pasiano da Bisticci school and attested in the letters. The original letters are archival
documents and manuscripts codices, held by cultural European institutions. All the
letters were transcribed, annotated and commented from philological, lexical, histori-
cal and prosopographycal points of view.
The purpose of the digital edition is to identify persons related to manuscripts, in
order to expose a datasets of people related to manuscripts, these ones described by
technical words. A first experiment of using HiCO (section 2.3) is a starting point for
exploring the potentiality of the ontology in this context, highlighting the network of
relations around the cultural object, such as a letter of the collection.
Fig. 4 Vespasiano’s Letters: the homepage of the website, an annotated letter,
the list of people’s authorities
3.3 Zeri Photo Archive
The Zeri Photo Archive, a rich digital catalog today considered one of the most im-
portant repertories of Italian art on the web, is an archival collection of photographs of
paintings stored in a traditional RDBMS.
We started a project for converting data to LOD by adopting a layered conceptual-
ization (namely, CIDOC-CRM) as both a descriptive and a conceptual model. We
first proceeded to reengineer the Entity/Relationship model provided by the database
tables, which structures data according to the Scheda F22 (Italian for F entry, a de-
scription standard issued by the Central Institute for Cataloguing and Documentation
[ICCD] for the cataloguing of photographic materials – where F stands for “Foto-
grafia”, photograph in Italian), into an OWL 2 DL ontology [5].
This was the first phase of a complete reconversion project, that will see the trans-
formation of the data currently stored in the database into RDF statements compliant
to a new ontology we are developing, and the use of automatic and semi-automatic
tools to generate links to existing datasets (fig. 5). The ontology itself is being itera-
tively enhanced following modifications of the ICCD Scheda F and of CIDOC-CRM,
making sure that the whole conceptual organization and entity naming of the existing
model are affected as little as possible.
22
Scheda F, http://www.iccd.beniculturali.it/index.php?it/387/beni-fotografici
Fig. 5 Zeri project workflow
4 The faceted browsing environment
The final idea is to use the potentiality of each ontology for exploring the semantic
content of all the cultural collections. The aim is to test if ontologies developed for a
specific domain could be suitable in other context. At the same time the idea is to
reveal possible connections, but also eventual contradictory statements, between data
in the collections. If an user searches for a person with a role, connected to a specific
place, and a specific date, or an aggregation of all (person, place and date) in an
event-oriented perspective, we assume we are able to understand the relationships
between the collections: i.e. the same person covering the same role in all the collec-
tions, or the same place described in all the collections with different functions.
In this regard, we adopt the term ‘facet’ both as a point of view on digital collec-
tions, i.e. as semantic lenses [10] for further clarifications on which quality standards
should data achieve, and a way for browsing results of search (categories and filters as
multi-level relationships).
In order to achieve these tasks the steps of the projects could be outlined in:
ontology matching, in order to verify a semantic integration – mostly needed be-
cause of the use of different names for describing the same concept and to under-
stand possible different conceptualizations for the same topic –, and to have a
common ontology for browsing all the datasets;
URI attribution, when needed, and normalization of the naming, i.e. URI design
related issues and identity resolution among datasets;
choice of shared authority files for people and places (es. LC authorities 23, VIAF24,
Geonames25, but also Freebase26 and Dbpedia27);
23
Library of Congress (LA) authorities, http://authorities.loc.gov/
24
Virtual International Authority File (VIAF), https://viaf.org/
XML markup enrichment, when needed, in order to let the ontologies able to ex-
tract information in a different context than the starting one;
definition of ad hoc query, starting from:
o people’s role detectable across collections;
o description of the same places and relationships involving them;
o definition of interpretations related to actions (someone do something)
and attributions (someone asserted something);
o events having in common a place, a date and/or a person;
selection of a faceted environment, in order to expose results of analysis and de-
ploy knowledge discovered [11].
5 Conclusions and future works
These steps are required to outline features of each dataset and to define which se-
mantic enrichment is needed in order to benefit of ontology matching: an accurate
analysis of the use cases will define more precisely when a model can be reused with
minimal efforts (mostly in ontology matching phase) and when a specific-domain
model should be considered only to address a restricted research topic. This will pro-
duce both an evaluation of ontologies and a refinement of higher-quality data.
Since first trials, it seems obvious that some specific-domain issues have to be
explicitly further formalized, representing a great effort in terms of semantic en-
hancement of starting data, e.g. attestation of roles where none are formally ex-
pressed.
As it is conceived, HiCO ontology could represent, without particular further
work, a superstructure to describe how places, events, roles and relations described in
datasets are bounded to cultural objects, dealing with provenance information and
avoiding contradictory statements will affect data consistency.
GO and PRoles, which are mainly devoted to describe above explained specific
issues, will be used, where possible, to enrich description of relations in some way
formalized but not still exploited for interrogation: e.g. where, in a latin text, a per-
son’s role is attested or could be deduced as strictly related to an ancient place, both
models can be used to accomplish the descriptive task and we will expect to benefit of
inferred information.
The expected result of data reorganization and enhancement will finally lead,
through the faceted search, to discover and generalize which requirements are needed
when creating datasets in a broad conceptualized perspective, allowing us to formal-
ize a shareable workflow for dealing with data related to cultural objects.
25
Geonames, http://www.geonames.org/
26
Freebase, https://www.freebase.com/
27
DbPedia, http://wiki.dbpedia.org/
6 Acknowledgments
This research is a result of a stimulating and fruitful collaboration on the single de-
scribed projects. We would like then to thank all people that worked with us on both
ontologies and digital collections: Ciro Gognano, Francesca Mambelli, Silvio Peroni,
and Fabio Vitali.
References
1. Lana, M., Ciotti, F., Magro, D., Peroni, S., Tomasi, F., Vitali, F.: Annotating texts with
ontologies, from geography to persons and events. Digital Humanities 2014,
http://dharchive.org/paper/DH2014/Poster-136.xml (2014)
2. Ciotti, F., Lana, M., Tomasi, F.: TEI, ontologies, linked open data: geolat and beyond.
Journal of the Text encoding Initiative, 8 (in print)
3. Daquino, M., Tomasi, F.: Ontological approaches to information description and extrac-
tion in the cultural heritage domain. In: Tomasi, F., Rosselli Del Turco, R., Tammaro,
A.M. (eds.) Humanities and Their Methods in the Digital Ecosystem. ACM, New York
(2015)
4. Daquino, M., Peroni S., Tomasi, F., Vitali F.: Political Roles Ontology (PRoles): enhanc-
ing archival authority records through Semantic Web technologies. Procedia Computer
Science 38, 60-67 (2014)
5. Gonano, C.M., Mambelli, F., Peroni, S., Tomasi, F., Vitali F.: Zeri e LODE. Extracting the
Zeri photo archive to Linked Open Data: formalizing the conceptual model. In: 14th
IEEE/ACM Joint Conference on Digital Libraries JCDL. IEEE, London (2014)
6. Lana, M.: Geolat: Geography for Latin Literature. In: Elliott, T., Heath, S., Muccigrosso,
J. (eds.) ISAW papers 7 Current Practice in Linked Open Data for the Ancient World,
http://sfsheath.github.io/lawdi-publication/isaw-papers-7.xhtml (2014)
7. Daquino, M., Tomasi, F.: Historical Context (HiCO): a conceptual model for describing
context information of cultural heritage objects. Communications in Computer and Infor-
mation Science 544, 424-436. Springer Verlag, Berlin (2015)
8. Tomasi, F.: L’edizione digitale e la rappresentazione della conoscenza. Un esempio: Ve-
spasiano da Bisticci e le sue lettere. Ecdotica 9, 264-286 (2013)
9. Mambelli, F.: Una risorsa online per la storia dell’arte: il database della fototeca Zeri. In:
Ciotti, F. (eds.) Digital Humanities: progetti italiani ed esperienze di convergenza multidi-
sciplinare. Quaderni Digilab, Università di Roma La Sapienza (2014)
10. Peroni, S., Tomasi, F., Vitali, F., Zingoni, J.: Semantic lenses as exploration method for
scholarly article. Communications in Computer and Information Science 385, 118-129.
Springer Verlag, Berlin (2014)
11. Pasin, M.: Browsing Highly Interconnected Humanities Databases Through Multi-Result
Faceted Browsers. Digital Humanities 2011, https://dh2011.stanford.edu/wp-
content/uploads/2011/05/DH2011_BookOfAbs.pdf (2011)