Concept-based Extension of SKOS Defense Controlled Vocabulary:
Techniques and Implications
PERICLES S. GIANNARIS1, NIKOLAOS DOUKAS2, NIKOS MASTORAKIS1
1Sector of Electrical Engineering and Computer Science,
ASEI (Military Institutes of University Education), Hellenic Naval Academy,
Hadjikyriakou Avenue, Piraeus, P.C. 18539,
GREECE
2Division of Mathematics and Engineering,
ASEI (Military Institutes of University Education), Hellenic Army Academy,
Leoforos Eyelpidon (Varis – Koropiou) Avenue Vari P.O. 16673,
GREECE
Abstract: - A controlled vocabulary is a set of terms that are utilized to represent knowledge in a domain. In the
domain of defense, the use of terms such as “command, control, communications, computers, intelligence,
surveillance and reconnaissance (C4ISR)”, “armored personnel carriers (APC)”, or “biological and bioinspired
structures for multispectral surveillance” denote the core competencies of domain experts and the depth of
diverse knowledge. This paper describes the second phase of the pilot project to create a defense-related
controlled vocabulary with a focus on the Russo-Ukraine conflict. Applications for data annotation, SKOS
hierarchical vocabulary development, and vocabulary quality analysis are used to identify terms in text, express
identified terms in military press releases as SKOS vocabulary, and assess its structure. The preliminary
vocabulary is extended by 173 concepts. The quality of the vocabulary is validated against a SKOS checker of
twenty-four criteria.
Key-Words: - Controlled Vocabulary, Knowledge Representation, SKOS, Semantic Web, Semantic Web,
Named Entity Recognition.
Received: June 19, 2023. Revised: February 8, 2024. Accepted: March 15, 2024. Published: May 7, 2024.
1 Introduction
A controlled vocabulary is a set of terms that are
utilized to represent knowledge in a domain. For
example, in the domain of defense are used the
following terms: “command, control,
communications, computers, intelligence,
surveillance and reconnaissance (C4ISR)”,
“armored personnel carriers (APC)”, or “biological
and bioinspired structures for multispectral
surveillance”. The structure that a controlled
vocabulary can acquire is that of a simple list of
terms or a complex graph structure. The World
Wide Web Consortium (W3C) considers controlled
vocabularies as the basic element of the Semantic
Web, a set of technologies to link disparate data
sources. To standardize the creation of a controlled
vocabulary in a machine-readable format and to
assist with its assessment, W3C recommends the
simple knowledge organization (SKOS) schema.
SKOS is based on the machine-readable resource
description framework (RDF) data model therefore,
it can be defined as an OWL ontology, [1], [2].
In the initial phase of our pilot study for the
creation of a Russo-Ukraine conflict controlled
vocabulary, we extracted terms from a sample of
North Atlantic Treaty Organization (NATO) press
releases. The terms in the vocabulary are
hierarchically structured according to the SKOS
data model. In this paper, we describe three steps to
extend our initial controlled vocabulary with new
concepts. First, we analyze a new batch of press
releases to identify terms to express in SKOS.
Second, we utilize the friend-of-a-friend (FOAF)
language to describe persons and their social links
within the context of NATO. FOAF depends on
W3C's standards, specifically on extensible markup
language (XML), XML Namespaces, RDF, and web
ontology language (OWL), [3]. Lastly, we conduct a
preliminary assessment of our extended controlled
vocabulary.
A couple of published research is noted to
demonstrate the utility of controlled vocabularies in
WSEAS TRANSACTIONS on INFORMATION SCIENCE and APPLICATIONS
DOI: 10.37394/23209.2024.21.22
Pericles S. Giannaris,
Nikolaos Doukas, Nikos Mastorakis
E-ISSN: 2224-3402
229
Volume 21, 2024
different disciplines. First, [4], emphasize that Gene
Ontology (GO) is a controlled vocabulary with three
hierarchies of terms for (a) gene functions, (b)
larger-scale biological processes, and (c) cellular
components to discover information about gene
products, [4]. Second, Based on a methodology to
semantically encode text data to machine
understandable format, [5], propose a controlled
vocabulary as the foundation of a thematic thesaurus
modelled according to SKOS to streamline the
interoperability of knowledge bases at knowledge-
intensive institutions such as libraries and
universities, [5].
The rest of the paper is organized as follows,
section two discusses the methodology for
extending our Russo-Ukraine conflict controlled
vocabulary based on text analysis and the
employment of tools to leverage SKOS and FOAF;
section three discusses pilot results, preliminary
evaluation of the controlled vocabulary and,
limitations; and section four concludes the paper.
2 Methodology
Figure 1 displays the tasks involved in creating a
controlled vocabulary for the Russo-Ukraine
conflict using terms from NATO press releases.
2.1 Data Acquisition
Similarly to the initial phase, we obtain press
releases from the NATO Press Office News site, [6].
Our query, “Ukraine AND Russia AND Conflict”,
covers the period from 27 July 2023 to 14
December, 2023. The search engine returns twenty-
eight press releases, between 23 August, 2023 and
14 December, 2023, in batches of ten per results
page. We manually store and organize the results in
a spreadsheet. The data are organized in three
columns: date, image, headline, and opening
paragraph.
2.2 Data Wrangling
This task has two steps. Firstly, we utilize the
"select object", "filter", "sort", "macro option for
Visual Basic for Application (VBA) code", and
"trim" functions in our spreadsheet editor to remove
images, standardize word spacing, convert hypertext
to plain text, and sort our sample of text data by
date. Our text data sample is now organized into
three columns: the date, the headline, and the press
release. Second, we measure the frequency of each
word in the data set and count data according to
unique dates, the number of press releases per date,
words per headline, and press release, [7], [8] using
R base functions, [9]. Text quantification provides a
preliminary view of the potential distribution of
terms considering our data set.
2.3 Text Data Annotation
Text data annotation is the process of tagging
information to text to circumstantiate its meaning.
As in the first phase of our research, we annotate
terms in our sample of press releases to be added
toour controlled vocabulary. Specifically, we
annotate text similarly to preparing data for named
entity recognition (NER), the computational
technique to recognize predefined categories of
entities in a body of text, [10] for example,
“[Secretary General]designation [Jens
Stoltenberg]proper_name visited [Kyiv]city”. To annotate
text, we employ Label Studio [11], a data annotation
tool for NER projects that is offered as a feature in
DagsHub [12], a platform that supports data
analytics projects. Additionally, we leverage the
code configuration interface of the tool to customize
the annotation tags to our previous manual
annotation codebook, [13]. The annotated data are
saved in a comma-separated values (CSV) file and
supplied in a JavaScript Object Notation (JSON)
like structure. Next, we programmatically parse the
data file to extract terms and their corresponding
annotations, [14], [15], in a data frame, a type of a
two-dimensional array, Table 1.
Table 1. Randomly selected terms and
corresponding annotation tags
Term
Annotation tag
nato deputy
secretary general
“designation”
vilnius summit
“defense_or_civil_event”
new york
“city”
intensive care units
in hospitals
“infrastructure”
italy
“country”
pressing security
challenges
“challenges_situation_problem”
artillery systems
“aegis_arms_research_instruments”
ministry of the
interior
“government_authority”
medical services to
civilians
“defense_process_measures”
WSEAS TRANSACTIONS on INFORMATION SCIENCE and APPLICATIONS
DOI: 10.37394/23209.2024.21.22
Pericles S. Giannaris,
Nikolaos Doukas, Nikos Mastorakis
E-ISSN: 2224-3402
230
Volume 21, 2024
Fig. 1: Flowchart of the process to create a Russo-Ukraine conflict SKOS-based controlled vocabulary
The data frame is stored in a spreadsheet for
further analysis. For each annotated term and
corresponding tag, we provide definitions and
related sources, which are used in the next task.
2.4 Creation of Controlled Vocabulary
SKOS has three basic elements that are defined as
classes, properties, and relations. Every element
begins with the prefix “skos”. The elements are
distinguished according to the uppercase or
lowercase letter after the prefix for example, a class
can be encoded as “skos:Concept”; a property can
be encoded as skos:prefLabel, [16]. The property
skos:related” can be used to express a relation
between two concepts. Importantly, SKOS models
data according to a concept-centered approach
compared to a term-centered approach, [17].
In this phase, we employ the Controlled
Vocabulary Editor (CoVEd), which uses the SKOS
schema and the Terse RDF Triple Language (TTL)
syntax to extend our controlled vocabulary. CoVEd
was developed by the Narralive research team of the
Athena Research and Innovation Center at the
University of Athens, Greece, [18], [19] in [20]. We
begin by creating a concept scheme that
incorporates the sets of concepts for our vocabulary.
Each concept is identified by a uniform resource
identifier (URI). The following SKOS concepts and
properties are the backbone of our vocabulary:
“skos:ConceptScheme”, “skos:Concept”,
“skos:inScheme”, “skos:topConceptOf”,
“skos:hasTopConcept”. The “skos:prefLabel”
encodes the preferred lexical label for our resource.
The use of the “skos:note” and “skos:definition”
properties helps us to provide information about our
SKOS concepts. Furthermore, the “skos:broader”
and “skos:narrower” properties are used to assert a
hierarchical connection between two concepts; the
“skos:related” property is used to declare semantic
relations between concepts in our concept scheme,
[21].
To encode data related to persons and their social
network in a semantic way, we have the opportunity
to use named properties and classes from the friend-
of-friend (FOAF) schema that are provided by
CoVEd. FOAF is based on RDF and the Web
Ontology Language (OWL) to link persons and
information using the Web, [22]. In this analysis, we
use the properties “foaf:person”, “foaf:name”,
“foaf:knows”.
3 Results and Discussion
Our search query returns 28 press releases published
by the NATO Press Office about the Russo-Ukraine
conflict between 27July, 2023, and 14December,
2023. The six most frequent words, as in unigram,
in our sample of press releases are the following:
“nato” 23 times, “secretary” 14 times, “Ukraine” 14
times, “general” 13 times, “support” 9 times, and
“defense” 6 times. We annotate 173 unique terms in
the text using twenty different tags, Table 2. The
annotation tags correspond to the hierarchy of
concepts in the SKOS vocabulary that we developed
during the first phase of this study.
WSEAS TRANSACTIONS on INFORMATION SCIENCE and APPLICATIONS
DOI: 10.37394/23209.2024.21.22
Pericles S. Giannaris,
Nikolaos Doukas, Nikos Mastorakis
E-ISSN: 2224-3402
231
Volume 21, 2024
Fig. 2: Snippet of the hierarchy of concepts of the
SKOS Russo-Ukraine Conflict vocabulary
As mentioned in the methodology, the result of
the NER-like annotation process is available in
JSON-based structure:
[{"end": 22, "text": "NATO Secretary
General", "start": 0, "labels":
["DESIGNATION"]}, [...], {"end": 413,
"text": "US-led Ukraine Defence Contact
Group meeting", "start": 369, "labels":
["DEFENCE_OR_CIVIL_EVENT"]}, [...],
{"end": 136, "text": "Brussels", "start":
128, "labels": ["CITY"]}, [...], {"end": 39,
"text": "Jens Stoltenberg", "start": 23,
"labels": ["PROPER_NAME"]}]
The above excerpt shows that JSON data are
structured as name-value pairs e.g., "text":
"Brussels" or "labels": ["CITY"]. Here, “text” is
paired with the term we identify in the press release;
“labels” is paired with the annotation tag we select
for that term.
Table 2. Excerpt of tags that are used to annotate
terms in press releases
Annotation tags
designation
proper_name
continent_or_state_or_province
defence_or_civil_event
nato_engagements
defense_project
biomedical
SKOS models data in concept schemes. Our SKOS
data structure has a five-level hierarchy of concept
specificity. The preferred label for our concept
scheme is expressed as ‘skos:prefLabel "russo-
ukrainian-defence-diplomacy-vocabulary"@en’.
SKOS Play, [23], prints the controlled vocabulary in
HTML format that allows to computationally
quantify the generated information. We count 224
concepts. Concerning terms, SKOS Play defines
“nato” as the top term (TT). Here, TT corresponds
to the above-mentioned first level of hierarchy. The
second level comprises twelve concepts; the third
level has eighteen concepts, and the fourth level has
twenty concepts. The above concepts are defined as
broad terms (BTs). The rest 173 concepts
correspond to terms identified in the sample of press
releases. The later are defined as narrow terms
(NTs). Figure 2 shows part of the tree structure of
our SKOS model. Furthermore, there are 180
relations expressed as “skos:related” between
concepts.
Below, excerpts of our SKOS model, in TTL
serialization, that models poly-hierarchy about the
concept “russell-c-leffingwell-lecture”:
ex4:nato rdf:type skos:Concept ;
dcterms:identifier
"http://example.com/periclesRepo/ex4/nato"^^xsd:anyURI ;
skos:inScheme ex4: ;
skos:topConceptOf ex4: ;
skos:narrower ex4:[. . .] ;
skos:narrower ex4:democratic_action ;
skos:narrower ex4:[. . .] ;
skos:prefLabel "nato"@en .
ex4:democratic_action rdf:type skos:Concept ;
dcterms:identifier
"http://example.com/periclesRepo/ex4/democratic_action"^
^xsd:anyURI ;
skos:broader ex4:nato ;
skos:inScheme ex4: ;
skos:narrower ex4:peace_building_event ;
skos:prefLabel "democratic_action"@en .
ex4:peace_building_event rdf:type skos:Concept ;
dcterms:identifier
"http://example.com/periclesRepo/ex4/peace_building_even
t"^^xsd:anyURI ;
skos:broader ex4:democratic_action ;
skos:inScheme ex4: ;
skos:narrower ex4:defence_or_civil_event ;
skos:prefLabel "peace_building_event"@en .
ex4:defence_or_civil_event rdf:type skos:Concept ;
WSEAS TRANSACTIONS on INFORMATION SCIENCE and APPLICATIONS
DOI: 10.37394/23209.2024.21.22
Pericles S. Giannaris,
Nikolaos Doukas, Nikos Mastorakis
E-ISSN: 2224-3402
232
Volume 21, 2024
dcterms:identifier
"http://example.com/periclesRepo/ex4/defence_or_civil_ev
ent"^^xsd:anyURI ;
skos:broader ex4:peace_building_event ;
skos:definition "Event: a thing that happens or takes place,
especially one of importance; a planned public or social
occasion"@en ;
skos:inScheme ex4: ;
skos:narrower ex4:[. . .] ;
skos:narrower ex4:russell-c-leffingwell-lecture ;
skos:narrower ex4:[]. . . ;
skos:note "source: https://languages.oup.com/google-
dictionary-en"@en ;
skos:prefLabel "defence_or_civil_event"@en .
ex4:russell-c-leffingwell-lecture rdf:type skos:Concept ;
dcterms:identifier
"http://example.com/periclesRepo/ex4/russell-c-leffingwell-
lecture"^^xsd:anyURI ;
skos:broader ex4:defence_or_civil_event ;
skos:definition "Russell C. Leffingwell Russell C.
Leffingwell Century Society Archives The Russell C.
Leffingwell Lecture, inaugurated in 1969, was named for a
charter member of CFR who served as its president from
1944 to 1946 and as its chairman from 1946 to 1953. This
lecture is given by a distinguished foreign official who is
invited to address CFR members on a topic of major
international significance. The lectureship was originally
endowed by the Morgan Guaranty Trust Company and by
Edward and Lucy Leffingwell Pulling, and more recently
through the generosity of Thomas Leffingwell Pulling and
his son Edward Leffingwell Pulling"@en ;
skos:inScheme ex4: ;
skos:note "source: https://www.cfr.org/project/russell-c-
leffingwell-lecture-series"@en ;
skos:prefLabel "russell c. leffingwell lecture"@en .
Below, example of relations between concepts
based on SKOS and FOAF in TTL serialization:
ex4:proper_name rdf:type skos:Concept ;
dcterms:identifier
"http://example.com/periclesRepo/ex4/proper_name"^^xsd:
anyURI ;
skos:broader ex4:person ;
skos:definition "A proper noun is the name of a particular
person, place, organization, or thing. Proper nouns begin
with a capital letter"@en ;
skos:inScheme ex4: ;
skos:narrower ex4:[. . .] ;
skos:narrower ex4:jens-stoltenberg ;
skos:narrower ex4:[. . .] ;
skos:note "source:
https://www.collinsdictionary.com/dictionary/english/prope
r-
noun#:~:text=A%20proper%20noun%20is%20the,Compare
%20common%20noun."@en ;
skos:prefLabel "proper_name"@en .
ex4:jens-stoltenberg rdf:type skos:Concept ;
dcterms:identifier
"http://example.com/periclesRepo/ex4/jens-
stoltenberg"^^xsd:anyURI ;
skos:broader ex4:proper_name ;
skos:definition "Jens Stoltenberg is a Norwegian politician
who has served as the 13th secretary general of NATO
since 2014"@en ;
skos:inScheme ex4: ;
skos:note "source:
https://en.wikipedia.org/wiki/Jens_Stoltenberg"@en ;
skos:prefLabel "jens stoltenberg"@en ;
skos:related ex4:[. . .] ;
skos:related ex4:joe-biden ;
skos:related ex4:[. . .] ;
foaf:knows ex4:[. . .] ;
foaf:knows ex4:joe-biden ;
foaf:knows ex4:[. . .] ;
foaf:name ex4:jens-stoltenberg ;
foaf:person ex4:jens-stoltenberg .
Interestingly, the Narralive research team
extends the CoVEd’s graph functionality with a
social network-like visualization for relations
between concepts expressed as “skos:related”,
Figure 3.
Fig. 3: The matrix represents extracts of relations between concepts. Relations are expressed as “skos:related”
WSEAS TRANSACTIONS on INFORMATION SCIENCE and APPLICATIONS
DOI: 10.37394/23209.2024.21.22
Pericles S. Giannaris,
Nikolaos Doukas, Nikos Mastorakis
E-ISSN: 2224-3402
233
Volume 21, 2024
The “graph” feature in CoVEd provides users
with an additional functionality to visualize skos
encoded relations in a social network-like depiction,
Figure 3. Here, “graph” is leveraged to illustrate
relations between concepts that are hierarchically
below “proper_name”. The nodes represent
concepts. The dotted threads link two concepts. (A)
Node that illustrates the higher level concept
“proper_name”; (B) and (C) Nodes that illustrate
narrower concepts, here, the proper names of NATO
leaders; (D) Link that illustrates relation between
two concepts. The network representation is used to
demonstrate potential professional relations between
individuals within the organizational context of the
Alliance.
3.1 Controlled Vocabulary Assessment
To assess the quality of our SKOS vocabulary, we
use the SKOS Play Testing Tool, which is frontend
for qSKOS, a tool for finding quality issues in
SKOS vocabularies, [24], [25], [26]. By default, are
selected twenty four rules that correspond to quality
checking functions to assess the vocabulary, [27].
The output is a report that states that 224 concepts
are processed. Of the 24 rules that are checked,
three failed to be verified. Two of the three failed
rules are warnings such as “uc - Undocumented
Concepts” and “urc - Unidirectionally Related
Concepts”. The warnings indicate that there are
thirty one concepts that are found not to use any
SKOS documentation properties, and two concepts
that do not include reciprocal relations. The one fail
found is the “var - Valueless Associative Relations”
rule. It refers to finding sibling concept pairs that are
also connected by an associative relation. The
following are excerpts from warnings:
uc:http://example.com/periclesRepo/ex4/democratic_action
urc:http://example.com/periclesRepo/ex4/farid-safarov
var:http://example.com/periclesRepo/ex4/jane-harman
This assessment of the vocabulary quality is an
initial step to determine its suitability for reuse and
extensibility in the development processes.
Furthermore, the checking functions flag potential
quality problems that can be interpreted as structural
errors. Or, they can hamper vocabulary integration
due to concept incompatibility and inconsistencies
3.2 Discussion
The preliminary text analysis of our sample of
NATO press releases provides an overview of the
volume of information pertaining to civil and
military affairs. Yet, text analysis is the groundwork
for capturing and structuring knowledge that could
be leveraged by computational models. Moreover,
text analysis aids in emphasizing the grammatical
form of the terms that we are interested in selecting.
For example, the vocabulary should contain terms in
one of the following grammatical forms: noun or
noun phrase for instance, “alliance” or “council on
foreign relations”; verbal noun for instance,
“hosting a multinational nato battlegroup”; adjective
or pre-modified phrase for instance, “humanitarian
aid”; and, post-modified noun phrase for instance,
“lessons learned on countering hybrid tactics” [28].
There are five levels in the hierarchical structure
of this SKOS vocabulary. Conceptually, the last
level corresponds to terms annotated during the
NER-like procedure. Our polyhierarchical system
for classifying terms in press releases is the source
of these levels. The report on the analysis of our
vocabulary that is generated by SKOS Play
distinguishes between broad terms (BT) and narrow
terms (NT) with “nato” being the top term (TT).
Generally, polyhierarchy denotes that concepts can
potentially belong to more than one category.
For relations between concepts, there are two
main types of SKOS properties: associative and
hierarchical. For example, the property
“skos:broader” asserts that a concept has a general
meaning and “skos:narrower” is the inverse
property. The following excerpt describes
hierarchical relations:
ex4: rdf:type skos:ConceptScheme ;
dcterms:identifier
"http://example.com/periclesRepo/ex4/"^^xsd:anyURI ;
skos:hasTopConcept ex4:nato ;
skos:prefLabel "russo-ukrainian-defence-diplomacy-
vocabulary"@en .
ex4:nato rdf:type skos:Concept ;
dcterms:identifier
"http://example.com/periclesRepo/ex4/nato"^^xsd:anyURI ;
skos:inScheme ex4: ;
skos:topConceptOf ex4: ;
skos:narrower ex4:[. . .] ;
skos:narrower ex4:democratic_action ;
skos:narrower ex4:[. . .] ;
skos:prefLabel "nato"@en .
ex4:democratic_action rdf:type skos:Concept ;
dcterms:identifier
"http://example.com/periclesRepo/ex4/democratic_action"^
^xsd:anyURI ;
skos:broader ex4:nato ;
skos:inScheme ex4: ;
skos:narrower ex4:peace_building_event ;
skos:prefLabel "democratic_action"@en .
ex4:peace_building_event rdf:type skos:Concept ;
dcterms:identifier
"http://example.com/periclesRepo/ex4/peace_building_even
t"^^xsd:anyURI ;
skos:broader ex4:democratic_action ;
skos:inScheme ex4: ;
skos:narrower ex4:defence_or_civil_event ;
skos:prefLabel "peace_building_event"@en .
ex4:defence_or_civil_event rdf:type skos:Concept ;
dcterms:identifier
"http://example.com/periclesRepo/ex4/defence_or_civil_ev
ent"^^xsd:anyURI ;
skos:broader ex4:peace_building_event ;
WSEAS TRANSACTIONS on INFORMATION SCIENCE and APPLICATIONS
DOI: 10.37394/23209.2024.21.22
Pericles S. Giannaris,
Nikolaos Doukas, Nikos Mastorakis
E-ISSN: 2224-3402
234
Volume 21, 2024
skos:definition "Event: a thing that happens or takes place,
especially one of importance; a planned public or social
occasion"@en ;
skos:inScheme ex4: ;
skos:narrower ex4:[. . .] ;
skos:narrower ex4:russell-c-leffingwell-lecture ;
skos:narrower ex4:[. . .] ;
skos:note "source: https://languages.oup.com/google-
dictionary-en"@en ;
skos:prefLabel "defence_or_civil_event"@en .
ex4:russell-c-leffingwell-lecture rdf:type skos:Concept ;
dcterms:identifier
"http://example.com/periclesRepo/ex4/russell-c-leffingwell-
lecture"^^xsd:anyURI ;
skos:broader ex4:defence_or_civil_event ;
skos:definition "Russell C. Leffingwell Russell C.
Leffingwell Century Society Archives The Russell C.
Leffingwell Lecture, inaugurated in 1969, was named for a
charter member of CFR who served as its president from
1944 to 1946 and as its chairman from 1946 to 1953. This
lecture is given by a distinguished foreign official who is
invited to address CFR members on a topic of major
international significance. The lectureship was originally
endowed by the Morgan Guaranty Trust Company and by
Edward and Lucy Leffingwell Pulling, and more recently
through the generosity of Thomas Leffingwell Pulling and
his son Edward Leffingwell Pulling"@en ;
skos:inScheme ex4: ;
skos:note "source: https://www.cfr.org/project/russell-c-
leffingwell-lecture-series"@en ;
skos:prefLabel "russell c. leffingwell lecture"@en .
The property “skos:related” expresses associative
connection between concepts. For example, we use
associative relations for concepts that are under the
proper_name category. However, according to the
W3C recommendation for SKOS vocabularies, it is
not required to encode associative relations for pairs
of concepts that share the same broader or narrower
concept categories, [28], [29]. Nevertheless, in this
phase, we intend to express links between
individuals that are associated with the alliance
under various professional capacities. Similarly, we
use the “foaf:name” and “foaf:knows” properties to
express relations between persons in the vocabulary.
The utilization of the properties “skos:definition”
and “skos:note” aims to provide a better
understanding of the meaning of a concept through
general documentation and the corresponding
source, [30]. Concerning limitations to the pilot
phase of our research, we underline two top issues.
First, the size of our data sample should be
increased to full press releases. Potentially, a larger
data set will provide additional terms for the
vocabulary. Consequently, it provides an increased
knowledge organization system (KOS). Second, the
polyhierarchy of categories of the SKOS vocabulary
is a structure that is under revaluation. The goal is to
address potential complicatedness regarding
broader-narrower concept relations. The next step is
to analyze entire press releases to expand the current
vocabulary.
Professional networks within the Treaty
connected by FOAF relationships can present
insights into characteristics and motifs of social
networks in the Semantic Web. Here, the use of
FOAF has the potential capacity to derive
organizational relations from contextual information
or domain knowledge through data mining
techniques such as classification clustering, or
logical inference.
4 Conclusion
This paper describes the second phase of the pilot
project to create a Russo-Ukraine conflict controlled
vocabulary. At this phase, CoVED and SKOS Play
Test are utilized to express 224 terms as SKOS
vocabulary and assess its structure. Of the twenty-
four rules that test the vocabulary, one rule is
marked as failed in the assessment report due to
Valueless Associative Relations. The analysis of full
press releases will be used as the next phase in this
research to potentially increase the number of
concepts in the vocabulary.The RDF-based
foundation of the proposed vocabulary provides a
nascent glimpse of an artificial intelligence (AI)
system to define classifications for defense
initiatives supporting peace-building civil and
diplomatic actions.
Acknowledgements:
We thank Dr. N. Mastorakis at the Hellenic Naval
Academy and Dr. N. Bardis at the Hellenic Army
Academy for facilitating this research.
Appreciations are also extended to Myrto Koukouli
and the NARRALIVE Research Team for providing
access to CoVEd and creating a repository for this
project.
References:
[1] A. Miles and S. Bechhofer, “SKOS Simple
Knowledge Organization System Reference,”
W3C Recommendation, [Online].
https://www.w3.org/TR/skos-reference/
(Accessed Date: February 15, 2024).
[2] I. Yoo and X. Hu, “Clustering Large
Collection of Biomedical Literature Based on
Ontology-Enriched Bipartite Graph
Representation and Mutual Refinement
Strategy,” in Advances in Knowledge
Discovery and Data Mining, in Lecture
Notes in Computer Science. Berlin,
WSEAS TRANSACTIONS on INFORMATION SCIENCE and APPLICATIONS
DOI: 10.37394/23209.2024.21.22
Pericles S. Giannaris,
Nikolaos Doukas, Nikos Mastorakis
E-ISSN: 2224-3402
235
Volume 21, 2024
Heidelberg: Springer, 2006, pp. 303–312,
https://doi.org/10.1007/11731139_36.
[3] D. Brickley and L. Miller, “FOAF
Vocabulary Specification,” FOAF
Vocabulary Specification, [Online].
http://xmlns.com/foaf/spec/ (Accessed Date:
February 15, 2024).
[4] J. H. Gennari, A. Silberfein, and J. C. Wiley,
“Integrating genomic knowledge sources
through an anatomy ontology,” Pac. Symp.
Biocomput. Pac. Symp. Biocomput.,
Fairmont Orchid on the Big Island of
Hawaii, USA, vol. 10, pp. 115–126, 2005,
ISBN 981-256-046-7.
[5] G. D. Solomou, D. A. Koutsomitropoulos, A.
K. Kalou, and S. D. Botsios, “Semantify
Educational Resources using SKOS and
Learning Object Ontologies,” 19th
International Conference on Circuits,
Systems, Communications and Computers
(CSCC 2015), Zakynthos Island, Greece, pp.
360–365, 2015.
[6] NATO Press Office, “NATO - News:
News,” NATO’s Newsroom, [Online].
https://www.nato.int/cps/en/natohq/news.htm
(Accessed Date: February 15, 2024).
[7] C. Khancome, “String Matching Algorithm
Using Multi-Characters Inverted Lists,”
WSEAS Transactions on Computers, Vol. 22,
pp. 151-158, 2023,
https://doi.org/10.37394/23205.2023.22.18.
[8] S. Oh, J. Jeong, C.-G. Lee, J. Yoo, and G.
Nam, “Synergistic Training: Harnessing
Active Learning and Pseudo-Labeling for
Enhanced Model Performance in Deep
Learning,” WSEAS Transactions on
Computers, Vol. 22, pp. 114-119, 2023,
https://doi.org/10.37394/23205.2023.22.14.
[9] The R Foundation, “R version 4.2.2 beta,” R:
The R Project for Statistical Computing,
[Online]. https://www.r-project.org/
(Accessed Date: February 15, 2024).
[10] IBM.com, “What is named entity
recognition?, IBM,” IBM, [Online].
https://www.ibm.com/topics/named-entity-
recognition (Accessed Date: February 15,
2024).
[11] Label Studio, “Label Studio Documentation
— Overview of Label Studio,” Label Studio,
[Online].
https://labelstud.io/guide/get_started.html
(Accessed Date: February 15, 2024).
[12] DagsHub, “DagsHub: The Home for
Machine Learning Collaboration,” DagsHub,
[Online]. https://dagshub.com/ (Accessed
Date: February 15, 2024).
[13] P. S. Giannaris, J. Borges, and N. Doukas,
“Design of a Defense Controlled
Vocabulary, SKOS-based Encoding of
Terms from Military Press Releases,” in
2023 13th International Conference on
Dependable Systems, Services and
Technologies (DESSERT), Athens, Greece,
Oct. 2023, pp. 1–6, doi:
10.1109/DESSERT61349.2023.10416475.
[14] S. Sivkov, “Residue Number Systems
Quantization for Deep Learning Inference,”
WSEAS Transactions on Computers, Vol. 22,
pp. 296-301, 2023,
https://doi.org/10.37394/23205.2023.22.33.
[15] S. Saranya and G. Usha, “Emotion
Classification on Social Media Comments
Using Categorical Feature Extraction Along
With the Bidirectional Encoder-based
Recurrent Neural Network Classification,”
WSEAS Transactions on Computers, Vol. 22,
pp. 284-295, 2023,
https://doi.org/10.37394/23205.2023.22.32.
[16] Linked Heritage, “Terminology,” Linked
Heritage, [Online].
http://linkedheritage.cab.unipd.it/training/LO
-07/en/02.html (Accessed Date: February 15,
2024).
[17] T. Baker, S. Bechhofer, A. Isaac, A. Miles,
G. Schreiber, and E. Summers, “Key choices
in the design of Simple Knowledge
Organization System (SKOS),” J. Web
Semant., vol. 20, pp. 35–49, May 2013, doi:
10.1016/j.websem.2013.05.001.
[18] Koukouli, M., Katifori, A., Roussou, M., &
Ioannidis, Y. (2024). Controlled Vocabulary
Editor. SKOS vocabularies without SKOS
training. In K. Karpouzis, Y. Skarpelos
(eds.), Computational Methods for the
Digital Humanities. Springer.
https://doi.org/10.5281/zenodo.11080329.
[19] M. Koukouli, A. Katifori, and Y. Ioannidis,
“SKOS vocabulary without SKOS training.
A tool for everyone,” presented at the
Symbiosis of Tradition and Digital
Technology, Tallinn, Estonia, May 22, 2022.
[20] CIDOC, “Symbiosis of Tradition and Digital
Technology CIDOC conference abstracts
MAY 23-26, 2022 Tallinn, Estonia.”
CIDOC, May 23, 2022. Accessed: Feb. 15,
2024, [Online].
https://vlaamsekunstcollectie.be/volumes/im
ported/Esinejate-abstraktid_0505.pdf
(Accessed Date: February 15, 2024).
WSEAS TRANSACTIONS on INFORMATION SCIENCE and APPLICATIONS
DOI: 10.37394/23209.2024.21.22
Pericles S. Giannaris,
Nikolaos Doukas, Nikos Mastorakis
E-ISSN: 2224-3402
236
Volume 21, 2024
[21] A. Miles and D. Brickley, “SKOS Core
Vocabulary Specification,” SKOS Core
Vocabulary Specification W3C Working
Draft 2 November 2005, [Online].
https://www.w3.org/TR/swbp-skos-core-
spec/#note (Accessed Date: November 19,
2024).
[22] D. Brickley and L. Miller, “FOAF
Vocabulary Specification,” FOAF
Vocabulary Specification 0.99, [Online].
http://xmlns.com/foaf/spec/ (Accessed Date:
February 15, 2024).
[23] T. Francart, “SKOS Play! - Thesaurus &
Taxonomies,” SKOS Play, [Online].
https://skos-play.sparna.fr/play/home
(Accessed Date: February 15, 2024).
[24] T. Francart and C. Maden, “SKOS testing
tool,” [Online]. https://skos-
play.sparna.fr/skos-testing-tool/ (Accessed
Date: February 15, 2024).
[25] C. Mader, “cmader/qSKOS.” Aug. 15, 2023,
[Online]. Available:
https://github.com/cmader/qSKOS (Accessed
Date: February 15, 2024).
[26] C. Mader, B. Haslhofer, and A. Isaac,
“Finding Quality Issues in SKOS
Vocabularies.” arXiv, Jun. 06, 2012.
[27] O. Suominen and C. Mader, “Assessing and
Improving the Quality of SKOS
Vocabularies,” J. Data Semant., vol. 3, no. 1,
pp. 47–73, Mar. 2014, 2024, doi:
10.1007/s13740-013-0026-0.
[28] I. Spasić, D. Schober, S. A. Sansone, D.
Rebholz-Schuhmann, D. B. Kell, N. W.
Paton, "Facilitating the development of
controlled vocabularies for metabolomics
technologies with text mining." BMC
Bioinformatics, 9 (Suppl 5), S5 (2008), pp.1-
16, https://doi.org/10.1186/1471-2105-9-S5-
S5.
[29] P. P. F. Barcelos, “Quality Issues,” GitHub,
[Online].
https://github.com/cmader/qSKOS/wiki/Qual
ity-Issues (Accessed Date: February 15,
2024).
[30] E. Summers, A. Isaac, C. Redding, D. Krech,
"LCSH, SKOS and Linked Data", Metadata
for Semantic and Social Applications, Dublin
Core: Innovation and Moving Forward,
DCMI '08: Proceedings of the 2008
International Conference on Dublin Core
and Metadata Applications, Berlin Germany
September 22-28, 2008, pp.25-33, [Online].
https://library.oapen.org/bitstream/handle/20.
500.12657/37029/1/DC_proceedings.pdf#pa
ge=41 (Accessed Date: April 26, 2024).
Contribution of Individual Authors to the
Creation of a Scientific Article (Ghostwriting
Policy):
The authors equally contributed tothe present
research, at all stages from the formulation of the
problem to the final findings and solution.
Sources of Funding for Research Presented in a
Scientific Article or Scientific Article Itself:
No funding was received for conducting this study.
Conflict of Interest:
The authors have no conflicts of interest to declare.
Creative Commons Attribution License 4.0
(Attribution 4.0 International, CC BY 4.0)
This article is published under the terms of the
Creative Commons Attribution License 4.0
https://creativecommons.org/licenses/by/4.0/deed.
en_US
WSEAS TRANSACTIONS on INFORMATION SCIENCE and APPLICATIONS
DOI: 10.37394/23209.2024.21.22
Pericles S. Giannaris,
Nikolaos Doukas, Nikos Mastorakis
E-ISSN: 2224-3402
237
Volume 21, 2024
