A Multilevel Graph-Based Recommender System for Personalized
Learning Paths in Archaeological Parks: Leveraging IoT and Situation
Awareness
Mario Casillo
1
, Francesco Colace
2
, Angelo Lorusso
2
, Domenico Santaniello
1
and Carmine Valentino
2
1
DISPAC, University of Salerno, Fisciano (SA), Italy
2
DIIN, University of Salerno, Fisciano (SA), Italy
Keywords:
Internet of Things, Bayesian Network, Situation Awareness, Ontology, Cultural Heritage.
Abstract:
Enhancing Cultural Heritage relies on innovative technologies to improve user interaction with cultural assets.
The advent of the Internet of Things (IoT) has made integrating smart devices with educational methodologies
possible, enabling a combination of cultural engagement, heritage promotion, and learning. This study aims to
introduce a Recommender System capable of suggesting personalized learning paths for users visiting archae-
ological parks, leveraging a multilevel graph-based approach. The method is grounded in Situation Awareness
(SA) and structured into three main levels: perception, comprehension, and prediction. The perception level is
ensured through data acquisition from sensors deployed in the field; the comprehension level utilizes seman-
tic and contextual graph approaches for domain representation; and the prediction level is developed using
predictive algorithms based on Bayesian Networks. A preliminary experimental campaign conducted across
three archaeological parks allowed for testing the effectiveness of the proposed approach, demonstrating its
predictive capabilities and potential in creating tailored cultural experiences. The findings highlight how ad-
vanced technologies can enrich users’ educational experiences and significantly contribute to the valorization
of cultural heritage.
1 INTRODUCTION
Enhancing cultural heritage requires employing novel
technologies to support users’ interactions with cul-
tural assets. The capability of providing personalized
experiences permits the improvement of enjoyment
(Hong et al., 2024) by suggesting appropriate tours
to users integrated with e-learning strategies. There-
fore, identifying systems that filter and analyze data to
understand user preferences is crucial (Casillo et al.,
2023; Colace et al., 2023).
Then, this work introduces the MuG Approach
(Casillo et al., 2024b), which is exploited as a Rec-
ommender System (RS) to suggest learning paths to
users visiting archaeological parks. In recent decades,
e-learning has been increasingly enriched with new
tools to improve the educational process. The typical
processes of the traditional education world, which
are still valid today, are assisted by the advent of
new technologies. The present era is characterized
by new intelligent devices capable of exchanging in-
formation with each other, contributing to the Inter-
net of Things (IoT) paradigm (Casillo et al., 2024a;
Michalakis and Caridakis, 2022). How can we take
advantage of such technologies to further enhance e-
learning? Suggesting training activities in a cultural-
historical center allows users to connect with histori-
cal assets, furthering the training process. However,
this is not enough to obtain good training results. It is
necessary to follow a well-structured educational path
provided by an expert guide whose objective is not
simply to describe the artifacts present as is typically
done with tourists but to give a historical-cultural per-
spective with a formative character (Jin et al., 2022).
This process is particularly complex. In this scenario,
having a methodology that deals with the automatic
design of educational paths could be interesting. This
technology could employ smart mobile devices and
the large amount of data they produce to build cus-
tomized learning paths. These training paths could
also be developed ad hoc concerning the archaeologi-
cal sites visited, allowing the combination of two ob-
jectives: promoting learning and enhancing cultural
heritage value.
The crucial point of the proposed work is achiev-
386
Casillo, M., Colace, F., Lorusso, A., Santaniello, D. and Valentino, C.
A Multilevel Graph-Based Recommender System for Personalized Learning Paths in Archaeological Parks: Leveraging IoT and Situation Awareness.
DOI: 10.5220/0013434500003944
Paper published under CC license (CC BY-NC-ND 4.0)
In Proceedings of the 10th International Conference on Internet of Things, Big Data and Security (IoTBDS 2025), pages 386-393
ISBN: 978-989-758-750-4; ISSN: 2184-4976
Proceedings Copyright © 2025 by SCITEPRESS – Science and Technology Publications, Lda.

ing Situation Awareness (SA). SA has emerged in
the literature as a central theme since the 1980s and
can be intuitively interpreted as awareness of what is
happening in a given context around people (Ends-
ley, 1995; Casillo et al., 2022; Stanton et al., 2001).
There is no single definition of SA, as different scien-
tific interpretations focus on distinct aspects: percep-
tion and understanding of the environment, interac-
tion between humans and the environment, and men-
tal models (Stanton et al., 2001). In general, SA is
concerned with human interaction, which is identifi-
able through tools such as projection, cognition, men-
tal models, perception and reflection, and the external
environment.
Endsley’s theory is based on a three-level ap-
proach (Endsley, 1995):
• Perception: acquisition of information about the
surrounding environment.
• Comprehension: processing information to be-
come aware of the environment.
• Prediction: using perceived and comprehension
information to predict future actions and antici-
pate problems.
The degree of awareness increases as we move be-
tween levels, reaching a maximum in the ability to
predict.
This work aims to achieve SA through three graph
approaches: the Multilevel Graph (MuG) approach
(Casillo et al., 2024b). Specifically, the level of per-
ception is achieved through the use of data acquired
through the IoT paradigm. Then, the Context Di-
mension Tree (CDT) and a Domain Ontology are ex-
ploited for the comprehension phase. Finally, the
structure of a Bayesian rate is identified via struc-
tural learning algorithms, which enables the predic-
tion phase.
Developing an experimental campaign for three
archaeological parks permits testing the MuG ap-
proach to suggest suitable user cultural experiences.
The campaign’s aim was to evaluate its predictive ca-
pabilities.
The paper is structured as follows: Section 2 fo-
cuses on state-of-the-art analysis, Section 3 describes
the proposed approach, Section 4 introduces the ex-
perimental campaign, and Section 5 presents conclu-
sions and future developments.
2 RELATED WORKS
Enhancement of cultural heritage by improving user
experiences is a topic of considerable interest in the
literature.
There are many techniques aiming to improve the
learning process. As Cicero liked to walk around
memorizing by associating physical paths to nar-
ration, new systems are increasingly aware of the
power of Digital Storytelling in e-learning. Digi-
tal Storytelling represents a traditional, modern take
on oral Storytelling, combining the ancient tradi-
tion of oral Storytelling with today’s technological
tools. There are many studies in the literature re-
garding the application of Digital Storytelling tech-
niques in educational pathways (Smeda et al., 2013;
Weng et al., 2011). In (Smeda et al., 2010), guide-
lines are proposed to develop an advanced frame-
work for e-learning that exploits the Digital Story-
telling technique. This is primarily by exploiting the
pedagogical-innovative capabilities of this approach
that has the potential in engagement that promotes im-
provement in learning.
Moreover, solutions to offer training models based
on Digital Storytelling to different groups of users
with different backgrounds and levels of digital liter-
acy are addressed in the literature to revive such learn-
ing models in developing countries (Ungerer, 2019).
This technique has been evaluated in different learn-
ing domains, including foreign language learning in
interdisciplinary projects obtaining exciting results
(Yang et al., 2022). Another fascinating field where
Storytelling is applied is in e-tourism and museums
to enhance Cultural Heritage (Chen et al., 2014; Ioan-
nidis et al., 2013).
Sometimes, Digital Storytelling approaches are
complemented with gamification. In (Rossano and
Roselli, 2018), a Content Management System for
Digital Storytelling to support knowledge acquisition
and fruition is proposed. This approach has obtained
interesting results in young patients with particular
health problems capable of influencing their emo-
tional sphere. Many studies in the literature propose
using e-learning systems that are able to use gamifi-
cation approaches (Amriani et al., 2014; Sanina et al.,
2020).
Such approaches aim to consolidate the training
path, using the capacity of modern technological sys-
tems, which, if well exploited, can adapt to users’
needs (Jianu and Vasilateanu, 2017). Based on this,
an interesting application is to combine gamification
with augmented reality to make the gaming experi-
ence more meaningful and enveloping (Bonsignore
et al., 2012). This approach has been found in many
areas, such as business production (Korn, 2012), so-
cial relations (Morschheuser et al., 2017), and e-
learning (Saidin et al., 2015). In particular, in (Pombo
et al., 2019) augmented reality has found excellent
feedback in learning paths especially outdoor.
A Multilevel Graph-Based Recommender System for Personalized Learning Paths in Archaeological Parks: Leveraging IoT and Situation
Awareness
387

Believing that the use of techniques as such is
valid in the e-learning field, another aspect to consider
is the use of such methodologies to enhance cultural
heritage. These methodologies are exploited to pro-
cess and interpret personal user information and con-
textual information. To this end, context can be used
to create applications (Dey, 2001; Raento et al., 2005)
that can filter relevant data by providing the correct
information at the right time and constantly updating
(Jin et al., 2014). Modern applications, in addition
to personal interests, can adapt to the user’s profile
(Fink and Kobsa, 2002), distinguishing, for example,
between a child and an adult, and can learn from for-
mer choices and provide real-time updates concerning
the context (Ghiani et al., 2009; Gavalas and Kenteris,
2011). Therefore, the need arises to create a method-
ology that combines the effectiveness of new techno-
logical devices to create training paths in the field.
The innovation of the proposed methodology is to ex-
ploit the capabilities of the new devices, the amount of
data they produce, and the REST services to automat-
ically design context-sensitive training paths valid for
different categories of users. These training paths, to
be performed in archaeological sites, address educa-
tion by collecting many innovative techniques in the
field of e-learning such as digital Storytelling, aug-
mented reality, and gamification.
This case study aims to use the MuG approach as a
recommendation system to suggest educational paths
exploiting a high degree of context awareness. This
approach is able to combine several methodologies
that underlie models working in different domains,
such as smart cities and cultural heritage enhance-
ment. In particular, in concurring with the prefixed
objective, that is, to recommend the right educational
path to the users according to the context, it is possible
to refer to the proposed methodology. This method
can bring semantic value to the available data to pro-
vide users with illustrated and augmented reality sto-
ries in proximity to the visited places and according
to different factors able to influence the educational
path such as available time, weather conditions, and
user’s attitudes.
3 THE PROPOSED APPROACH
This section describes the proposed approach to
achieving situation awareness by employing three
graphs. It also aims to describe how the proposed ap-
proach is applied to enhance cultural heritage. This
task requires the definition of an architecture based
on four functional layers to provide personalized ser-
vices to users consisting of e-learning cultural paths
chosen appropriately for users.
The section is divided into two subsections: the
first describes the MuG approach, while the second
contextualizes it for improving user experience.
3.1 The Multilevel Graph Approach
The Multilevel Graph (MuG) approach aims to make
the architecture that will be described in subsection
3.2 achieve Situation Awareness. To do this, it is nec-
essary to achieve the three levels described in the in-
troductory section: perception, via data acquired from
sensors and API Services, understanding, and predic-
tion.
Underlying the proposed approach are three
graphs: the Context Dimension Tree (CDT) and Do-
main Ontology that enable the comprehension stage,
and the Bayesian Network (BN) that enables the pre-
diction stage.
The CDT is a context representation model based
on a graph
G
c
=< r
c
, N
c
, E
c
>, (1)
specifically a tree, with a set of nodes (N
c
) and arcs
(E
c
) (Bolchini et al., 2009). The nodes are divided
into:
• Dimension nodes: represent dimensions, shown
in black.
• Concept nodes: represent dimension values,
shown in white.
• Attribute nodes: represent attributes and are al-
ways leaves of the tree.
Specifically, the root node r
c
is a concept node rep-
resenting the most general context. Dimension nodes
have only concept nodes as children, attribute nodes
can have only dimension or concept nodes as their
parents and cannot have children, and each attribute
node is a unique child. Dimension nodes without con-
cept children must have at least one attribute child.
The CDT alternates between dimension nodes and
concept nodes, forming distinct generations. Each
node is characterized by its type (dimension or con-
cept) and a unique label determined by the path con-
necting it to the root. Finally, parameters can be asso-
ciated with leaf nodes to further refine data selection.
Figure 1: Example of Context Dimension Tree.
IoTBDS 2025 - 10th International Conference on Internet of Things, Big Data and Security
388

Ontology is a valuable tool for representing,
sharing, and reusing information (Gruber, 1995).
It reduces conceptual and terminological confusion
through a shared vocabulary that includes precise
definitions and semantic relationships between enti-
ties. This formalization enables clear communica-
tion between humans and machines, ensuring that
each piece of information has a precise meaning
related to its context. An ontology is a formal
and explicit specification of a shared conceptualiza-
tion. It includes concepts, attributes, and hierar-
chical and semantic relationships formalized to be
machine-processable. Ontologies can be classified
into lightweight (more straightforward, with essential
relationships) and heavyweight (more complex, with
additional axioms and constraints).
Let formalize the Ontology
O =< C , A, H, R
+
, R
−
>, (2)
where C are the concepts, A the attributes, H the hi-
erarchical relations, and R
+
and R
−
dependency and
independency relations. As with the CDT, the Ontol-
ogy can be represented by a graph
G
0
=< r
0
, N
0
, E
0
>, (3)
where the set of nodes N
O
consists of the concepts
nodes and the set of edges
E
0
=

(n
1
, λ, n
2
) : n
1
, n
2
∈ N
0
, λ ∈ H ∪ R
+
∪ R
−

(4)
presents labels that describe the typology of connec-
tions among concepts. In addition, dimensions and
concepts nodes of the CDT are a subset of the con-
cepts C of the Ontology and attribute nodes of the
CDT are a subset of the attributes A of the Ontology.
The interaction between CDT and the Ontol-
ogy allows the definition of constraints to improve
the knowledge of the dependency and independency
among nodes. These constraints permits to improve
the construction of the structure of the BN. Specifi-
cally, the BN can be represented by the graph
G
B
=< N
B
, E
B
>, (5)
where the nodes N
B
are a subset of the concepts of
the Ontology and the edges represent the dependency
connections among the nodes N
B
.
Specifically, Algorithm 1 describe the structural
learning procedure integrated with the constrains
identified by the interaction between the Ontology
and the CDT.
Specifically, the algorithm requires the nodes N
B
of the BN, the dependency and independency con-
straints D and I, perspectively, and the definition of
the order of nodes and a maximum number of parents
u. In addition, the algorithm requires the definition of
a function g(·, ·) that allows to evaluate the reliability
level of identified BN (see (Casillo et al., 2024b)).
Algorithm 1: Algorithm of the MuG approach.
Data: N
B
, dependency constraints D,
independency constraints I, node
order, max number of connections u.
Result: printout of the node’s relatives for
each node.
for h = 1, . . . , n do
¯
π
h
←
{
(X, X
i
) ∈ D
}
⊆ D;
P
old
← g (h,
¯
π
h
);
status ← True;
while status AND
|
¯
π
h
|
≤ u do
Let be z ∈ Prec (X
h
) and
¯
π
h
which
maximizes g(h,
¯
π
h
∪
{
z
}
);
P
new
:= g (h,
¯
π
h
∪
{
z
}
);
if Pnew > P
old
then
P
old
← P
new
;
¯
π
h
←
¯
π
h
∪
{
z
}
;
else
status ← False;
end
end
write: Node X
h
- Parents
¯
π
h
end
3.2 The Proposed Architecture
After describing the MuG approach, we want to de-
scribe the architecture enabling this methodology to
enhance users’ cultural experiences and cultural her-
itage.
The architecture, depicted in Figure 2, has four
layers: the Acquisition Layer, the Knowledge-Base
Layer, the Inference Engine Layer, and the Applica-
tion Layer.
The first layer aims at data acquisition through
sensors and API services. The central point of this
layer is the IoT paradigm. The ability to manage de-
vices that not only acquire data of interest but are also
capable of sending this data via the Internet to cen-
tralizers that allow it to be managed and sent to Cloud
platforms allows for data to be available that not only
describes the context in which the system acts but also
allows for the improvement of the suggestions made
by acquiring data about both the users and the points
of interest that constitute the paths that will be sug-
gested. This layer exploits sensors capable of acquir-
ing data related to meteorological conditions and user
location. In addition, data acquired through external
sources is integrated via API services.
After the data acquisition phase, the Knowledge-
Base Layer comes into play. At this stage, the raw
data needs a preprocessing and cleaning phase that
allows storage in the database. The stored data pop-
A Multilevel Graph-Based Recommender System for Personalized Learning Paths in Archaeological Parks: Leveraging IoT and Situation
Awareness
389

Figure 2: The proposed architecture for enhancing cultural
heritage based on IoT and the MuG approach.
ulates the system, allowing it to move from the per-
ception to the comprehension phase by employing
CDT and Domain Ontology. The CDT allows us to
understand the context in which the system is act-
ing, while the Ontology allows us to gain knowledge
about the data through the defined relationships. Us-
ing these two graphs completes the comprehension
phase and allows for identifying the BN structure
that will enable the prediction phase. Specifically,
the Knowledge-Base Layer performs the online phase
that enables training by integrating structural learning
techniques with the constraints defined through the in-
teraction between CDT and Ontology (Colace et al.,
2019). In addition, this phase is reproduced periodi-
cally to allow the prediction capability to be improved
periodically due to new data acquired from the archi-
tecture.
The Inference Engine Layer, on the other hand,
exploits the online phase of the MuG approach
through the Prediction Module to figure out the best
path to suggest to the user. Similarly, the Digital
Storytelling Module is leveraged to identify multime-
dia content that enables the implementation of story-
telling strategies aimed at user engagement. Finally,
the Experience Elaboration Module enables the inte-
gration of Prediction Module predictions with Digital
Storytelling Module processing.
Finally, the Application Layer permits users to ac-
cess elaborated services, which consist of suggested
paths and related multimedia content.
4 EXPERIMENTAL RESULTS
The evaluation process of the proposed methodology,
in this case study, was conducted through the devel-
opment of a prototype based on the proposed archi-
tecture. The prototype consists of a server compo-
nent and a hybrid mobile app. The technologies used
were the Ibernate framework, based on Java, to build
the Rest API server-side service; the Apache Cordova
Framework for developing the mobile app.
The experimental phase was performed in three
archaeological parks in southern Italy: The Archaeo-
logical Park of Paestum, the Archaeological Park of
Herculaneum, and the Archaeological Park of Pom-
peii. Even if the proposed architecture allows the
System to access several open data on the web, the
prototype has been modeled with particular attention
to the three archaeological sites in this first experi-
mental phase. Several training modules related to the
considered archaeological parks have been inserted
inside the System, allowing the System to build the
related educational paths. The different training mod-
ules have been inserted in order to collect the needs
of different users (children, adults, and experts) dur-
ing the training path.
A total of 230 users were involved, trying to di-
vide them homogeneously by different age groups
and characteristics, who were unaware of the purpose
of the research. Each participant was equipped with
the application prototype, and they enjoyed different
training modules within the specific archaeological
parks. Users were divided into three different groups
composed as follows:
• Group 1: Archaeological Park of Pompeii (95
users)
• Group 2: Paestum Archaeological Park (65 users)
• Group 3: Archaeological Park of Herculaneum
(70 users)
IoTBDS 2025 - 10th International Conference on Internet of Things, Big Data and Security
390

Following the experience of using the content, a ques-
tionnaire divided into the following sections was pro-
posed to all users:
A. Presentation
1. The information has been presented appropri-
ately.
2. The information provided was exhaustive.
B. Reliability
1. The System provided path suggestions during
the entire visit.
2. The System was able to run adequately during
the whole visit.
C. Recommendation
1. The proposed services and contents have sat-
isfied the needs of the user, based on personal
preferences and the current context.
2. The System has managed to adapt to context
changes.
D. Performance
1. The System was able to show smooth operation
and without unexpected jumps.
2. Response times are adequate.
E. Usability
1. The system interface is user-friendly.
2. The System is able to provide suggestions with-
out being unwelcome
Based on the Likert scale, each section of the
questionnaire had two assertions to which five possi-
ble responses were associated: totally disagree - TD,
disagree - D, Undecided - U, agree - A, totally agree
- TA. The responses were collected in Table 1
Table 1: Case of Study on paths recommender: Question-
naire answers.
Sections Answer
TD D U A TA
A 18 21 89 187 145
B 20 27 55 230 128
C 18 3 26 239 174
D 11 65 82 203 99
E 37 28 46 205 144
In addition, a smaller number of participants were
asked to participate in the experimental phase to eval-
uate the System’s effectiveness in suggesting services.
To this end, five training modules were selected for
each archaeological park, and users were allowed to
indicate whether such a proposed module was rele-
vant according to their needs and context. The partic-
ipants who took part in the experimental phase for the
second time are divided as follows:
• Group 1: Archaeological Park of Pompeii (43
users)
• Group 2: Paestum Archaeological Park (32 users)
• Group 3: Archaeological Park of Herculaneum
(34 users)
In this experimental phase, the System’s knowl-
edge base was augmented by the data that emerged
from the experience of previous users. The results,
expressing the relevance of the proposed training
modules to the context and needs, were collected in
the form of a confusion matrix (Figure 4(a), Figure
4(b), Figure 4(c)).
Table 1 shows the degree of satisfaction of the 230
participants. Users agree that the System is able to
provide training modules that are tailored and in line
with the context.
In Figure 3, the results obtained are shown graph-
ically. Users are most satisfied with the ability to rec-
ommend the right training path concerning the con-
text. The confusion matrices shown in Figure 4(a),
Figure 4(b), Figure 4(c) show that the System was
able to recommend suitable training modules to users
based on the profile and time requirements of the
users. All the confusion matrices report an overall ac-
curacy greater than 70%, very encouraging data. Fig-
ure 4(a) brings back an overall accuracy advanced to
85%; this extraordinary result can be due to two fac-
tors. A factor could be the choice of the formative
modules that turn out particularly adapted to the se-
lected place. The second factor could be related to
the size of the Archaeological Park of Paestum. Un-
like other sites, due to its medium-large size and the
layout of the archaeological finds, it is better suited to
itinerant training and augmented reality. However, all
the results obtained are very encouraging and could
improve over time by the users’ experiences.
The objective of this case study was to validate the
proposed methodology in supporting users to choose
training paths within archaeological parks. The aim
was to provide tailored training content making the
training experience adaptable to the context and the
user’s needs. This case study confirms that the pro-
posed architecture could be declined in different con-
texts and mobile applications. The experimental re-
sults are promising and encouraging; they show that
the System is able to design training paths effectively
and that the developed prototype is efficient from sev-
eral points of view. The degree of reliability the us-
ability of the prototype have been evaluated very pos-
itively by the users involved in the experimental cam-
paign. In addition, the recommendation ability of the
System reached a high level of accuracy.
A Multilevel Graph-Based Recommender System for Personalized Learning Paths in Archaeological Parks: Leveraging IoT and Situation
Awareness
391

Figure 3: Case of Study on paths recommender: Trend of questionnaire answers.
(a) Case of Study on E-Learning paths recom-
mender: Confusion Matrix Group 1
(b) Case of Study on E-Learning paths recom-
mender: Confusion Matrix Group 2
(c) Case of Study on E-Learning paths recom-
mender: Confusion Matrix Group 3
Figure 4: Confusion matrices related to Group 1, Group 2,
and Group 3.
5 CONCLUSIONS AND FUTURE
WORKS
The article aims to exploit the MuG approach as a
Recommendation System to improve the enjoyment
of cultural heritage. Through the Internet of Things
paradigm and using three graphs (CDT, Onotlogy, and
Bayesian Network), it was possible to suggest person-
alized experiences to users while exploiting Digital
Storytelling techniques to build e-learning paths. The
experimental phase presented confirms the goodness
of the proposed approach. Future developments con-
cern the integration of XR strategies, such as Aug-
mented Reality, Virtual Reality, or Mixed Reality, to
better engage users in the cultural experience. In ad-
dition, further future evaluations will address the po-
tential for generalization of the proposed approach
through application to additional case studies outside
the world of cultural heritage.
REFERENCES
Amriani, A., Aji, A. F., Utomo, A. Y., and Junus, K. M.
(2014). An empirical study of gamification impact on
e-learning environment. In ICCSNT 2013, page 265 –
269.
Bolchini, C., Curino, C., Quintarelli, E., Schreiber, F., and
Tanca, L. (2009). Context information for knowledge
reshaping. International Journal of Web Engineering
and Technology, 5(1):88 – 103.
Bonsignore, E. M., Hansen, D. L., Toups, Z. O., Nacke,
L. E., Salter, A., and Lutters, W. (2012). Mixed reality
games. In CSCW, page 7 – 8.
Casillo, M., Colace, F., Gaeta, R., Lorusso, A., Santaniello,
D., and Valentino, C. (2024a). Revolutionizing cul-
tural heritage preservation: an innovative iot-based
framework for protecting historical buildings. Evol.
Intell., 17(5-6):3815 – 3831.
Casillo, M., Colace, F., Gupta, B. B., Lorusso, A.,
Marongiu, F., Santaniello, D., and Valentino, C.
(2022). A situation awareness approach for smart
home management. In ISMODE 2021, page 260 –
265.
Casillo, M., Colace, F., Gupta, B. B., Lorusso, A., San-
taniello, D., and Valentino, C. (2023). The role of
AI in improving interaction with cultural heritage: An
overview. IGI Global Scientific Publishing.
IoTBDS 2025 - 10th International Conference on Internet of Things, Big Data and Security
392

Casillo, M., Colace, F., Lorusso, A., Santaniello, D., and
Valentino, C. (2024b). A multilevel graph approach
for iot-based complex scenario management through
situation awareness and semantic approaches. Reliab.
Intell. Environ., 10(4):395 – 411.
Chen, H.-C., Kao, Y.-F., and Kuo, C.-L. (2014). A multi-
media storytelling in a rural village: The show taiwan
e-tourism service using tablet technologies. In IIAI-
AAI 2014, page 525 – 526.
Colace, F., Conte, D., Gupta, B., Santaniello, D., Troiano,
A., and Valentino, C. (2023). A novel context-aware
recommendation approach based on tensor decom-
position. Lecture Notes in Networks and Systems,
448:453 – 462.
Colace, F., Lombardi, M., Pascale, F., and Santaniello, D.
(2019). A multilevel graph representation for big data
interpretation in real scenarios. In ICSRS 2018, page
40 – 47.
Dey, A. K. (2001). Understanding and using context. Pers.
Ubiquitous Comput., 5(1):4 – 7.
Endsley, M. (1995). Toward a theory of situation awareness
in dynamic systems. Hum. Factors, 37(1):32 – 64.
Fink, J. and Kobsa, A. (2002). User modeling for personal-
ized city tours. Artif. Intell. Rev., 18(1):33 – 74.
Gavalas, D. and Kenteris, M. (2011). A web-based
pervasive recommendation system for mobile tourist
guides. Pers. Ubiquitous Comput., 15(7):759 – 770.
Ghiani, G., Patern
`
o, F., Santoro, C., and Spano, L. D.
(2009). Ubicicero: A location-aware, multi-device
museum guide. Interact. Comput., 21(4):288 – 303.
Gruber, T. R. (1995). Toward principles for the design of
ontologies used for knowledge sharing. Int. J. Hum.
Comput. Stud., 43(5-6):907 – 928.
Hong, M., An, S., Akerkar, R., Camacho, D., and Jung,
J. J. (2024). Cross-cultural contextualisation for rec-
ommender systems. J. Ambient Intell. Humaniz.,
15(2):1659 – 1670.
Ioannidis, Y., Raheb, K. E., Toli, E., Katifori, A., Boile, M.,
and Mazura, M. (2013). One object many stories: In-
troducing ict in museums and collections through dig-
ital storytelling. In Proceedings of the DigitalHeritage
2013, volume 1, page 421 – 424.
Jianu, E. M. and Vasilateanu, A. (2017). Designing of an e-
learning system using adaptivity and gamification. In
ISSE 2017.
Jin, J., Gubbi, J., Marusic, S., and Palaniswami, M. (2014).
An information framework for creating a smart city
through internet of things. IEEE Internet Things J.,
1(2):112 – 121.
Jin, S., Fan, M., and Kadir, A. (2022). Immersive spring
morning in the han palace: Learning traditional chi-
nese art via virtual reality and multi-touch tabletop.
Int. J. Hum. Comput., 38(3):213 – 226.
Korn, O. (2012). Industrial playgrounds: How gamification
helps to enrich work for elderly or impaired persons
in production. In EICS’12, page 313 – 316.
Michalakis, K. and Caridakis, G. (2022). Context awareness
in cultural heritage applications: A survey. J. Comput.
Cult. Herit., 15(2).
Morschheuser, B., Riar, M., Hamari, J., and Maedche, A.
(2017). How games induce cooperation? a study
on the relationship between game features and we-
intentions in an augmented reality game. Comput.
Hum. Behav., 77:169 – 183.
Pombo, L., Marques, M. M., Afonso, L., Dias, P., and
Madeira, J. (2019). Evaluation of a mobile augmented
reality game application as an outdoor learning tool.
Int. J. Mob. Blended Learn., 11(4):59 – 78.
Raento, M., Oulasvirta, A., Petit, R., and Toivonen, H.
(2005). Contextphone: A prototyping platform for
context-aware mobile applications. IEEE Pervasive
Comput., 4(2):51 – 59.
Rossano, V. and Roselli, T. (2018). Game-based learning
as effective learning method: an application of digital
storytelling. In IV 2018, page 542 – 546.
Saidin, N. F., Halim, N. D. A., and Yahaya, N. (2015). A
review of research on augmented reality in education:
Advantages and applications. Int. Educ. Stud., 8(13):1
– 8.
Sanina, A., Kutergina, E., and Balashov, A. (2020). The co-
creative approach to digital simulation games in social
science education. Comput. Educ., 149.
Smeda, N., Dakich, E., and Sharda, N. (2010). Developing
a framework for advancing e-learning through digital
storytelling. In MCCSIS 2010, volume 1, page 169 –
176.
Smeda, N., Dakich, E., and Sharda, N. (2013). The effec-
tiveness of digital storytelling in the classrooms: A
case study. In ICALT 2013, page 491 – 492.
Stanton, N., Chambers, P., and Piggott, J. (2001). Situa-
tional awareness and safety. Saf. Sci., 39(3):189 – 204.
Ungerer, L. (2019). Digital storytelling: Possible applica-
tions in an open distance e-learning environment. At
the Interface: Probing the Boundaries, 122:74 – 87.
Weng, J.-F., Kuo, H.-L., and Tseng, S.-S. (2011). Interac-
tive storytelling for elementary school nature science
education. In ICALT 2011, page 336 – 338.
Yang, Y.-T. C., Chen, Y.-C., and Hung, H.-T. (2022). Dig-
ital storytelling as an interdisciplinary project to im-
prove students’ english speaking and creative think-
ing. Comput. Assist. Lang. Learn., 35(4):840 – 862.
A Multilevel Graph-Based Recommender System for Personalized Learning Paths in Archaeological Parks: Leveraging IoT and Situation
Awareness
393