Machine Learning applications in
bibliometrics. Scopus bibliometric analysis of consumer behavior
Aplicaciones del Machine Lear=
ning
en bibliometría. Análisis bibliométrico Scopus =
del
comportamiento del consumidor.
|
1 |
Emilio Enrique Guerra Castell=
ón |
|
https://orcid.org/0009-00=
05-2436-7186 |
|
|
|
Investigador independiente |
|||
|
2 |
Investigador independiente |
|||
|
3 |
Investigador independiente |
![](3D"02paperok.files/image004.png")
|
|
Artículo de Investigación Científica y Tecnológica Enviado: 11/07/2024 Revisado: 19/08/2024 Aceptado: 10/09/2024 Publicado:05/10/2024 DOI: =
https://doi.org/10.33262/ct.v3i=
4.51
=
<=
/span> |
||||||||
|
|
|
||||||||
|
Cítese: |
|
Guer=
ra
Castellón, E. E., Núñez Torres, E., & Vázquez Alfonso, Y. (2024).
Aplicaciones del Machine Learning en bibliome=
tría.
Análisis bibliométrico Scopus del comportamie=
nto
del consumidor. Ciencia & Turismo, 3(4), 23-40. https://d=
oi.org/10.33262/ct.v3i4.51 |
|
||||||
|
|
|
La revista es editada por la Editorial Ciencia
Digital (Editorial de prestigio registrada en la Cámara Ecuatoriana de Li=
bro
con No de Afiliación 663) www.celibro.org.ec |
|
||||||
|
|
|
Esta
revista está protegida bajo una licencia Creative Co=
mmons
AttributionNonCommercialNoDerivatives 4.0
International. Copia de la licencia: http://creativecommons.org/licenses/by-nc-nd/=
4.0/ |
|
||||||
|
Palabras claves: Análisis bibliométrico,
machine Learning, comportamiento del consumid=
or,
aplicaciones, bibliometría. |
|
Resumen Este artículo explora las
aplicaciones del aprendizaje automático en bibliometría, específicamente
aplicando este conocimiento a un estudio bibliométrico del comportamiento=
del
consumidor. El crecimiento exponencial de las publicaciones científicas ha
superado las capacidades de las herramientas bibliométricas tradicionales,
creando la necesidad de técnicas más avanzadas para analizar grandes
cantidades de datos. El aprendizaje automático, una rama de la inteligenc=
ia
artificial, permite a las computadoras aprender de los datos y hacer
predicciones sin instrucciones explícitas. El estudio utiliza una metodol=
ogía
bibliométrica descriptiva, utilizando 2563 artículos de Scopus
y algoritmos de aprendizaje automático como Latent
Dirichlet Allocation (LDA) para la clasificac=
ión de
artículos. Se identificaron cinco temas principales: investigación de
mercado, influencia en línea en el consumo de alimentos y la percepción d=
e la
marca, implicaciones sociales del consumo, implicaciones para la salud y =
la
relación entre el comportamiento de compra y el marketing. Los resultados
revelan una gran red de colaboración internacional con Estados Unidos y el
Reino Unido a la cabeza y autores clave como Rifkin, Almotairi,
Gruber, Kunz, |
|
||||||
|
Keywords: Bibliometric
analysis, machine Learning, consumer behaviour,
applications, bibliometrics. |
|
Abstract This article explores the applications of machine
learning in bibliometrics, specifically by applying this knowledge to a
bibliometric study of consumer behaviour. The
exponential growth of scientific publications has outstripped the
capabilities of traditional bibliometric tools, creating a need for more
advanced techniques to analyse large amounts =
of
data. Machine learning, a branch of artificial intelligence, allows compu=
ters
to learn from data and make predictions without explicit instruction. The
study uses a descriptive bibliometric methodology, using 2563 articles fr=
om
Scopus and machine learning algorithms such as Latent Dirichlet Allocation
(LDA) for article classification. Five main topics were identified: market
research, online influence on food consumption and brand perception, soci=
al
implications of consumption, health implications, and the relationship
between shopping behaviour and marketing. The
results reveal a large international collaborative network with the US an=
d UK
leading the way and key authors including Rifkin, Al=
motairi,
Gruber, Kunz, Bonnevie, Buil, Hieke, Lal, Wang and Carlson. Machine learn=
ing
is emerging as a key tool for bibliometrics, allowing processes to be
automated, complex patterns to be identified and the efficiency of analys=
is
of large databases to be improved. This innovative approach facilitates t=
he
understanding of trends and the development of knowledge about consumer <=
span
class=3DSpellE>behaviour. |
|
||||||
![](3D"02paperok.files/image006.png")
![3Deditorial1.png](3D"02paperok.files/image008.jpg")
![](3D"02paperok.files/image010.jpg")
<=
span
lang=3DEN-US style=3D'font-size:12.0pt;line-height:115%;font-family:"Times =
New Roman",serif;
mso-fareast-font-family:"Times New Roman"'>=
1. Introduction
Bibliometric analysis has established itself as an
essential tool for studying the dynamics of scientific production and the
evolution of knowledge (Guerra et al., 2024). However, the exponenti=
al
growth in the volume of scientific publications has outstripped the
capabilities of traditional bibliometric tools, creating a need for more
advanced techniques to manage and analyse large
volumes of data.
Machine learning, a subfield of artificial
intelligence (AI), focuses on developing algorithms that allow computers to
learn from data, make predictions and make decisions without specific
instructions (Sarker, 2021). These capabilities are beneficial in bibliomet=
ric
analysis, where the volume of information to be processed requires automati=
on
and advanced accuracy.
The integration of machine learning into scientometrics has revolutionized the way researchers
manage and explore large databases. This approach has made it possible to
automate key processes such as document classification, the identification =
of
emerging trends and the analysis of scientific collaboration networks,
improving efficiency and the ability to identify patterns that were previou=
sly
undetectable.
In recent years, there has been a remarkable incre=
ase
in the scientific production of bibliometric analyses related to consumer <=
span
class=3DSpellE>behaviour, driven by the interest in understanding th=
is
area from a quantitative perspective. Although several techniques and
computational tools have been used, the application of machine learning has=
not
yet been extensively explored, which represents an opportunity to overcome =
the
current limitations in identifying complex and emerging patterns.
Important studies, such as Saravia et al.
(2022), have examined consumer behaviour betwee=
n 2011
and 2021. This analysis is carried out in the Scopus database and is notable
for its comprehensiveness, identifying 1,258 relevant documents. They used R
Studio software with the Bibliometrix package. =
They
note that the most emerging topics related to the subject matter are sales,
decision-making, trade, surveys, purchasing, e-commerce, sustainable
development, structural equation modelling and the theory of planned behaviour. Pérez et al. (2021) focus specifica=
lly on
tourism consumer behaviour. They present a
bibliometric analysis of articles published between 2009 and 2018 in journa=
ls
indexed in Scopus, including a sample of 330 articles. They used EndNote X11
bibliographic manager, Excel, VOSviewer r 1.6.8=
and
NVIVO 12 for data processing. They highlight that the most productive autho=
rs
according to Lotka's law are Johan Bruwer and R=
ob
Law. The results of the content analysis showed that there are five main no=
des
into which the research can be grouped: consumer behav=
iour,
travel, marketing, destination and experience, where topics related to the
purchase decision process, brand loyalty, use of technology and sustainabil=
ity
are addressed. The intersection between neuroscience and consumer behaviour is also an area of growing interest. Barrer=
a et
al. (2022) conducted a bibliometric analysis of papers published between
2007 and 2021, identifying and analyzing three research perspectives:
advertising and its emotional impact; brands and their quest to persuade
consumers; emotions and their influence on the brain and consumer behaviour. In addition, Camarena and Romero's (2023) =
study
on sustainable and responsible consumption, through a bibliometric analysis=
of
articles published between 2000 and 2020, reveals an increase in scientific
production on these concepts and highlights a paradigm shift from responsib=
le
consumption to sustainable consumption.
Given the growing volume of scientific data and the
need to optimize bibliometric analysis, this research is justified in that machine learning offers advanced tools to meet=
these
new demands. While previous studies have applied traditional bibliometric
analysis to consumer behaviour, none have explo=
red
the potential of machine learning to automate document classification, iden=
tify
emerging patterns and improve the efficiency of large database analysis.
Therefore, this study aims to fill this gap by first reviewing the main
applications of machine learning applied to bibliometric analysis of consum=
er behaviour from 2014 to 2024, based on the Scopus data=
base.
Machine Learning. Review of applications in
bibliometrics
Today, we live in the data age, where everything
around us is connected to a data source and everything is digitally recorded
(Cao, 2020). For example, today's electronic world has a
large number of different types of data, such as Internet of Things
(IoT) data, cybersecurity data, smart city data, commercial data, smartphone
data, social media data, health data, bibliographic data, and many more. Da=
ta
can be structured, semi-structured or unstructured.
Data types and their role in bibliometrics
Structured data is data that is organized in rows =
and
columns to facilitate processing, searching and analysis. Typical examples =
are
databases that store records in tables, such as those generated by bibliome=
tric
platforms such as Scopus, Dimensions and Web of Science. These data are ide=
al
for processing by supervised learning algorithms such as decision trees, ra=
ndom
forest and gradient boosting, thanks to their compatibility with libraries =
and
tools such as Pandas, NumPy, Scikit-learn and TensorFlow.
On the other hand, semi-structured data, such as J=
SON
or XML, which have a flexible structure, and unstructured data, such as free
text or PDF documents, present greater analysis challenges due to their
variability in format and content. However, artificial intelligence (AI) and
machine learning (ML) have evolved to efficiently process this data, allowi=
ng
applications to organize, analyse and automatic=
ally
learn from the information (Sarker et al., 2021).
Machine learning and its application in bibliometr=
ics
Machine learning, in particula=
r, has
emerged as a powerful tool for bibliometric analysis. Several techniques th=
at
can be divided into supervised, unsupervised, semi-supervised and reinforce=
ment
learning (Mohammed et al., 2016). In addition, deep learning, a rece=
nt
and popular technique, has shown great potential in identifying complex
patterns in large amounts of data (Janiesch, 20=
21).
Its main applications in bibliometrics include:
(1) docum=
ent
classification, i.e. automatically assigning articles to specific categorie=
s or
topics. The supervised learning algorithms used, as in Junior et al.
(2012), are Naïve Bayes, k-nearest neighbours (=
k-NN)
and support vector machine (SVM), which were used in a clustering method to
achieve greater accuracy, precision and coverage than traditional methods
alone, and to categorize production engineering articles efficiently and
automatically. Another such method is random forest, which was used by Garc=
ia
(2018) to perform sensitivity analysis in bibliometric models. This approach
allows the importance of different variables in the classification of docum=
ents
to be assessed, using simulated and real data. The research highlights how =
the
algorithm improves model accuracy by avoiding problems such as overfitting,
which is crucial when analyzing large amounts of data. Unsupervised learning
algorithms used for this purpose are Latent Dirichlet Assignment (LDA) in
studies such as Urbizagastegui (2021), which an=
alyses
Brazilian bibliometrics. The author notes that, when faced with the situati=
on
common to researchers of finding a huge collection of documents, the first
impulse is to organize and group them. To do this, they use this algorithm,
which uses complex mathematics and helps in the processes of automatic text
classification. With the help of RStudio, they were able to classify the
literature into five themes with the most recurrent words and to identify t=
he
authors.
(2) Document clustering, which, unlike classificat=
ion,
groups articles according to their thematic similarity without using labels.
Algorithms such as K-means and Latent Dirichlet Allocation (LDA) have been =
used
in Fuentealba (2023), in which they obtained 19 clusters on the literature
associated with VUCA, validating them with the coherence index and silhouet=
te,
respectively. Other techniques include density-based spatial clustering of
applications with noise, K-medoids and hierarchical clustering (Vasquez =
et
al., 2019).
(3) Citation prediction is another of the most
relevant applications of machine learning, as it allows us to know how many
citations an article will receive in the future based on historical data, w=
hich
is very relevant for researchers to know which high-impact, current and
relevant topics to choose to develop their research. Methods such as linear
regression (Cabeza, 2019), and the random forest to predict the number of
citations that articles in the field of biomedicine will receive, where
features such as the number of co-authors, journal quality and keywords were
used to train the model (Medina and Ñique, 2017=
), and
others such as neural network models (Neural Network, =
TabNet)
and LightGBM, the latter identified by Hu et=
al.
(2023) as having the highest performance.
(4) Co-authorship network analysis to detect group=
s or
communities of collaborating authors. The main algorithms most
commonly used in the literature are community detection and the Louv=
ain
method, both applied in bibliometric studies (Carratal=
á
et al., 2012; Knop et al., 2019), which also use visualization
techniques to map collaborative networks between researchers, revealing
patterns of co-authorship and highlighting the most influential communities=
in
the field.
(5) Automatic screening or sifting of the literatu=
re
using deep or active learning algorithms (Forero and B=
ennasar,
2022), speeding up reviews to find the most relevant research quickly, with=
out
having to spend several days reading. Following this idea, one of the tools
that has gained popularity is ASReview Lab, whi=
ch
uses state-of-the-art active learning techniques to solve one of the most
interesting challenges in the systematic selection of large amounts of text=
(ASReview, 2024).
The use of machine learning in bibliometrics has
advanced the analysis and organization of large volumes of scientific data =
and
will continue to evolve in the coming years. The processing of structured d=
ata
offers a significant advantage for the use of supervised techniques, while
unstructured data requires more complex techniques, such as deep learning, =
for
efficient analysis. However, the effective use of these techniques requires
well-organized datasets, appropriate pre-processing and an optimal choice o=
f ML
algorithms depending on the type of data and the research objectives.
<=
span
lang=3DEN-US style=3D'font-size:12.0pt;line-height:115%;font-family:"Times =
New Roman",serif;
mso-fareast-font-family:"Times New Roman"'>=
2. Methodology
For this article, a literature review was conducte=
d to
qualitatively analyse the applications of machi=
ne
learning in bibliometrics, and a descriptive bibliometric study was conduct=
ed.
From the point of view of Cohen and Gómez (2022), it consists of identifyin=
g,
summarizing, systematizing and analyzing all the information generated in
relevant information sources to maintain a timely and effective relationship
with the subject under analysis. Therefore, to develop the article, a revie=
w of
the articles published in the Scopus databases was carried out. In addition=
to
this, the study has a quantitative approach, through bibliometric numbers, =
to
make a comparison of the evolution and contribution of the perspectives of
different authors on the subject, and the research modality is bibliographic
since it uses secondary data as a source of information (Torres, 2016). To
proceed with the research, (1) the scope of the research was determined, wh=
ich
includes two aspects: a) as context, full-text publications and b) publicat=
ions
of articles related to studies with the keyword "consumer behaviour"; (2) the research questions were defi=
ned,
which appear in Table 1.
Table
1
Research
questions.
|
Research questions. |
|
1.=
What
is the trend in article publications and citations per year? |
|
2.=
What
are the topics of the articles? |
|
3.=
Who
are the most representative authors for each topic? |
|
4.=
What
are the main collaborative networks between authors and countries? |
|
5.=
Which
are the most cited articles? |
Source: Own
elaboration.
Subsequently, (3)
inclusion and exclusion criteria were defined. The inclusion criteria deter=
mine
which articles are selected for the research. The exclusion criteria limit =
the
articles that will not be part of the analysis. The inclusion and exclusion=
criteria
used are described in Table 2 below.
Table 2
Inclusion and exclusion criteria.
|
Inclusion criteria |
Exclusion criteria |
|
1. Journal
articles published between 2014 and 2024. 2. Articles
containing the keyword: ‘consumer behaviour’. 3.=
Open
access articles. |
1. Articles
in languages other than English. 2. Articles
from systematic reviews or bibliometric analysis. 3.=
Duplicate
articles. |
Source: Own elaboration.
Next, 2563 articles that met the inclusion criteria
were processed. The machine learning algorithms were modelled using R Studio
software, in this case, Latent Dirichlet Allocation (LDA), and the bibliometrix package (Aria and Cuccurullo, 2017) and
ggplot2 were used to visualize the graphs (Wickman, 2016).
<=
span
lang=3DEN-US style=3D'font-size:12.0pt;line-height:115%;font-family:"Times =
New Roman",serif;
mso-fareast-font-family:"Times New Roman"'>=
3. Results
The analysis shows an increasing trend in scientif=
ic
output from 2014 to 2024, with a steady increase in the number of published
articles, peaking in 2022 with 396 publications. However, the number of
citations does not follow a homogeneous pattern over time. Despite the
increased production in recent years, citations show a significant decrease
from 2020 onwards, which could indicate that more recent articles have not =
had
enough time to be widely cited or that the visibility and impact of research
published in recent years on this topic has decreased. This behaviour
could also be related to increased competitiveness or the need to dissemina=
te
more recent work more widely. The significant drop in citations for 2023 and
2024 is particularly noteworthy, with 2292 and 593 citations respectively,
highlighting the importance of continuously monitoring these indicators over
time to assess their academic impact.
Figure
1
The
Trend in article production and citations.
![](3D"02paperok.files/image012.png")
Source: Own
elaboration.
Figure 2 shows the
distribution of the number of articles for each of the five topics identifi=
ed
by the Latent Dirichlet Allocation (LDA) machine learning algorithm. There =
is
considerable variability in the number of publications associated with each
category. Topics 1 and 5 concentrate the largest number of articles,
significantly outnumbering the other categories. In contrast, themes 3 and 4
have a lower frequency of publications. This uneven distribution suggests a
concentration of scientific production in Themes 1 and 5, which could indic=
ate
a greater relevance or research interest in these specific areas within the
field of study in question.
Figure 2
Distribution of articles by topic.
![](3D"02paperok.files/image014.jpg")
=
Source: Own el=
aboration.
Figure 3 shows the
probability distribution of the most relevant terms within each of the five
topics identified by the LDA model for consumer behavi=
our
studies. Each topic represents an underlying theme in the analyzed document=
s,
and the most likely terms within each topic provide a window into their
semantic content. There is a clear thematic distinction between the themes.=
For
example, Topic 1 focuses on market research, which uses data collection to
explore variables of perceived product quality that influence consumer
purchases. Topic 2 brings together research on the influence of the online
environment on food consumption and brand perception. Theme 3 addresses iss=
ues
related to the social impact of food consumption, as well as predictive
modelling and analysis of food consumption data. Topic 4 focuses on the imp=
act
of specific foods on health and consumer behaviour.
Theme 5 hosts research that examines consumer behaviou=
r
and its relationship to marketing and product purchase. Figure 3 Top 10 words by topic. Source: Own
elaboration. The bibliometric anal=
ysis
of the relevant authors in the five themes shows a differentiated distribut=
ion
of citations, reflecting the influence and recognition in their respective
fields. In Theme 1, the most prominent authors are Rifkin and Almotairi, with 442 and 435 citations respectively,
indicating their pre-eminence in this field. On the other hand, in Topic 2,
Gruber leads with 1635 citations, followed by Kunz (1205), indicating a
significant impact on the related literature. Topic 3 is dominated by Bonne=
vie
(750 citations) and Buil (366), while in Topic 4 Hieke is the most cited au=
thor
with 849 citations, followed by Lal (368). In Topic 5, Wang (967 citations)=
and
Carlson (904) stand out as the most influential authors. These data show a
strong concentration of citations on certain authors, suggesting academic
leadership in their respective topics. Table 3 Top authors in each field. Authors Topic Cites Authors Topic Cites Authors Topic Cites Rifkin 1 442 Gruber 2 1635 Bonnevie 3 750 Almotairi 1 435 Kunz 2 1205 Buil 3 366 Petit 1 290 Scarpi 2 580 Ben Ismail 3 344 Dwivedi 1 281 Pantano 2 555 Liao 3 298 Goh 1 279 Dennis 2 476 Ernest 3 190 Table 3 Top authors in each field (continuation) Authors Topic Cites Authors Topic Cites Authors Topic Cites Hieke 4 849 Wang 5 967 Lal 4 368 Carlson 5 904 Moeller 4 283 Su 5 321 Segovia 4 268 Brad 5 277 Savva 4 200 Hingorani 5 273 Source: Own
elaboration. Figure 4 shows the
dispersion of the number of citations by each of the five themes identified=
by
the LDA model, color-coded. Each dot represents an author and its position =
on
the horizontal axis indicates the dominant topic, while the vertical axis r=
eflects
the total number of citations received. There is considerable variation in =
the
number of citations between authors. This variation suggests that the autho=
rs
assigned to Themes 2 and 5, such as Gruber, Kunz, Wang, Carlson, Hieke and =
Scarpi, are the most influential and have had the gre=
atest
impact on the scientific community in terms of citations received. Figure 4 Distribution of citations of authors in each field=
. Source: Own
elaboration. The analysis of
scientific collaboration between countries reveals an interconnected network
that highlights strong collaboration between some specific geographical
regions. In particular, countries such as the Un=
ited
States and the United Kingdom show a high frequency of collaboration and a =
high
scientific output (40 collaborations), suggesting a consolidated scientific
relationship. Other countries, such as Australia and the United Kingdom (26=
collaborations),
the United Kingdom and Italy (24 collaborations), and the United Kingdom and
China (23 collaborations), also show significant collaborative links. These
data reflect the geopolitical and academic influence in building scientific
networks, as well as the importance of consumer behavi=
our
at an international level, with countries with high levels of investment in
R&D leading international collaboration. The analysis also suggests that
there are opportunities to expand collaborations in regions where the frequ=
ency
is low or emerging, as in the case of Australia and Brazil (2 collaboration=
s).
These results allow us not only to understand the distribution of internati=
onal
scientific output but also to identify potential gaps and strategic areas f=
or
strengthening global scientific cooperation. Figure 5 Collaborations between countries Source: Own el=
aboration The analysis of the
academic collaboration network reveals a clear distribution of authors in
different clusters, with key metrics reflecting their importance within the
network. The Betweenness metric highlights authors such as Zhang Y, with a
value of 77, indicating his crucial role as a facilitator of collaboration
between different groups of researchers. In contrast, authors such as Chen S
and Dwivedi YK have Betweenness values of 0, suggesting less mediation in t=
he
network. The closeness metric, which reflects an author's proximity to the =
rest
of the network, shows that Wang Y has the highest value (0.02), followed by
Zhang Y (0.027), which places them in a position accessible to the rest of =
the
researchers. In terms of 'PageRank', which measures overall influence, Wang=
Y
leads with a value of 0.065, while Zhang Y has a value of 0.049. These resu=
lts
show the importance of certain authors, such as Wang Y and Zhang Y, both in
terms of mediation and overall influence in the network, positioning them a=
s key
players in the collaborative development of the field. Figure 6 Network of collaboration between authors. Source: Own
elaboration. The most cited articl=
es
are listed in Table 4. The data highlights the impact and relevance of each=
article
in its field. The article with the highest number of citations is Wirtz (20=
18)
with 1205 citations, although it is the second article in terms of citations
per year (172.14), indicating a sustained impact, but not necessarily the m=
ost
recent. The paper by Dwivedi (2021) stands out with the highest citation ra=
te
per year (226.00), suggesting rapid recognition in the scientific community.
Other articles, such as those by Grunert (2014) and Zhu (2014), although ol=
der,
maintain a considerable citation rate (77.18 and 68.18), reflecting their
relevance in their fields over time. Table 4 Most cited articles Artículo DOI Cites C per year WIRTZ J, 2018, J SERV MANAGE=
1205 172.14 DWIVEDI YK, 2021, INT J INF MAN=
AGE 904 226.00 GRUNERT KG, 2014, FOOD POLICY 849 77.18 ZHU SH, 2014, TOB CONTROL 750 68.18 Table 4 Most cited articles (continuation) Artículo DOI Cites C per year PANTANO=
E,
2020, J BUS RES 476 95.20 NAPOLI J, 2014, J BUS RES 447 40.64 KAMBOJ S, 2018, INT J INF MANAG=
E 444 63.43 KIRK CP, 2020, J BUS RES 442 88.40 BENOIT S, 2017, J BUS RES 430 53.75 DWIVEDI YK, 2016, GOV INF Q<=
o:p> 368 40.89 Source: Own elaborati=
on. 4.&n=
bsp;
Conclusion ·&nb=
sp;
Machine
learning is presented as an essential tool for modern bibliometric analysis.
The volume of scientific publications has grown exponentially, making it
difficult to analyse with traditional tools, and
allowing the automation of complex processes that previously required a gre=
at
deal of effort from researchers. ·&nb=
sp;
There
are several machine learning techniques that have been successfully applied=
to
bibliometrics. Supervised, unsupervised and reinforcement machine learning,=
as
well as deep learning, have proven their usefulness. Applications such as
document classification, clustering, citation prediction, collaborative
networks and automatic literature screening stand out. ·&nb=
sp;
Bibliometric
analysis of consumer behaviour benefits from ma=
chine
learning. The study shows an increasing trend in scientific production on t=
his
topic but with a decrease in citations from 2020 onwards. The use of the LDA
algorithm allowed the identification of 5 main themes, with key authors suc=
h as
Rifkin, Almotairi, Gruber, Kunz, Bonnevie, Buil,
Hieke, Lal, Wang and Carlson leading research in their respective areas. ·&nb=
sp;
International
scientific collaboration in the field of consumer beha=
viour
is robust but with potential for expansion. Analysis of cross-country
collaborative networks shows strong links between the US, UK, Australia, It=
aly
and China, reflecting their influence on research. Opportunities to strengt=
hen
collaboration with other regions are identified. The network of collaborati=
ons
between authors shows the importance of figures such as Wang and Zhang as
facilitators and influencers. Analysis of the most cited articles highlights
the impact of work such as that of Wirtz and Dwivedi. =
5. Conflict
of interest The authors declare t=
hat
there is no conflict of interest. =
6. Author
contribution statement All authors contribut=
ed
significantly to the preparation of the article. =
7. Financing
costs This research was fun=
ded
entirely with the authors' own funds 8.
References Aria, M. y Cuccuru=
llo,
C. (2017). Bibliometrix: An R-tool for comprehensive science mapping analysis. Journal o=
f Informetrics, 11(4), 959-975. https://doi.org/10.1016/=
j.joi.2017.08.007 ASReview. (2024, 11 de Octubre=
). Aprendizaje activo para
revisiones sistemáticas. https://asreview.nl/ Barrera Rodríguez,=
A.
M., Duque Hurtado, P. L., y Merchán Villegas, V. L. (2022). Neurociencia y
comportamiento del consumidor: análisis estadístico de su evolución y
tendencias en su investigación. Cuadernos Latinoamericanos De
Administración, 18(35). =
https://doi.org/10.18270/cuaderlam.v18i35.3855 Cabeza Ramírez, L.=
J.
(2019). Aproximación bibliométrica a la investigación en emprendimiento<=
/i>
[Tesis Doctoral]. Universidad de Córdoba. =
https://helvia.uco.es/bitstream/handle/10396/18690/2019000001933.pdf?sequen=
ce=3D1&isAllowed=3Dy Camarena Gómez, B. O., y
Romero Valenzuela, D. M. (2023). El consumo sustentable y responsable:
conceptos y análisis desde el comportamiento del consumidor. Revista Vértice Universitario, 25(=
94). https://d=
oi.org/10.36792/rvu.v25i94.75 Cao, L. (2020). Data Science: A Comprehensive Overview. ACM
Computing Surveys, 50(3), 43:1-42. http://dx.doi.org/10.114=
5/0000000.0000000 Carratalá Munuera,=
M.
C., Orozco Beltrán, D., Gil Guillen, V. F., Navarro Pérez, J., Quirce, F.,
Merino, J., y Basora, J. (2012). Análisis Bibliométrico de la producción
científica internacional sobre atención primaria. Atención Primaria,=
44(11),
651-658. =
https://doi.org/10.1016/j.aprim.2011.12.002 Cohen, N; y Gómez, R. (202=
2).
Pensar el para qué de la Metodología de la investigación. Un recorrido por =
la
producción de los datos y los diseños. Revista
Latinoamericana de Metodología de las Ciencias Sociales, 12(2), e122. <=
/span>https://d=
oi.org/10.24215/18537863e122 Forero Corba,
W., y Bennasar, F. N. (2024). Técnicas y aplica=
ciones
del Machine Learning e Inteligencia Artificial =
en
educación: Una revisión sistemática. RIED-Revista
Iberoamericana de Educación a Distancia, 27(1). https://www.redalyc.org/journal/3314/331475280017/html/ Fuentealba, D.,
Flores-Fernández, C., y Carrasco, R. (2023). Análisis bibliométrico y de
contenido sobre VUCA. Revista Española de Documentación Científica, =
46
(2), e354. https://doi.org/10.3989=
/redc.2023.2.1968 García, M. G. (2018). Anál=
isis
de sensibilidad mediante Ramdom Forest [Tesis de
Ingeniería]. Universidad Politécnica de Madrid. https://oa.upm.es/53368/1/TFG_MARTA_GARCIA_RUIZ_DE_LEON.pdf Guerra Castellón, =
E.
E., Sotero Arteaga, L. K., Vázquez Alfonso, Y., y Colcha Ortiz, A. R. (2024=
).
Diagnóstico de la producción científica registrada sobre la transformación
digital en el turismo a partir de un análisis bibliométrico. AlfaPublica=
ciones,
6(2.1), 46–63. https://doi.org/10.3326=
2/ap.v6i2.1.476 Hu, Z., Cui, J., y
Lin, A. (2023). Identif=
ying
potentially excellent publications using a citation-based Machine Learning
approach. Information Processing & Management, 60(3), 103=
323. https://doi.org/10.1016/j.ipm.2023.103323 Janiesch, C., Zschech,=
P., y
Heinrich, K. (2021). Machine Learning and deep learning. Electronic Mark=
ets,
31(3), 685-695. https://doi.org/10.1007/s12525-021-00475-2 Junior, A., Gomes, F., y Rodrigues, G. (20=
12).
Categorização
automática de artigos científicos da engenharia de produção utilizando méto=
dos
de aprendizagem de máquina. Encontro Nacional De Engenharia De
Produção: Desenvolvimento Sustentável E Responsabilidade Social: As
Contribuições da Engenharia de Produção, 32. https://abepro.org.br/bi=
blioteca/enegep2012_tn_sto_164_956_20299.pdf Knop, S., Merchel, R., y Poeppelbuss,
J. (2019). Author Collaboration in Ten Years of IPS2: A Bibliometric Analys=
is. Procedia CIRP, 83, 22-27=
. ![](3D"02paperok.files/image016.jpg")
=
![](3D"02paperok.files/image018.png")
=
![](3D"02paperok.files/image022.jpg")
=
![](3D"02paperok.files/image024.png")
=
Medina Merino, R. =
F.,
y Ñique Chacón, C. I. (2017). Bosques aleatorios como extensión de los árbo=
les
de clasificación con los programas R y Python. Interfases, =
;10(010),
165-189. =
https://doi.org/10.26439/interfases2017.n10.1775
Mohammed, M., Khan=
, M.
B., y Bashier, E. B. M. (2016). Machine Learning: Algorithms and Applications. CRC Press. https://doi.org/10.1201=
/9781315371658
Pérez Ricardo, E. =
C.,
Batista Sánchez, E., Medina Labrada, J. R., y Fernández Cueria, N. (2021).
Comportamiento del consumidor turístico. Análisis bibliométrico en Scopus.
Pensar la Publicidad. Pensar la publicidad, 15(2), 289-298. <=
/span>=
https://doi.org/10.5209/pepu.77526
Sailema Altamirano, J. C., Vargas
Lizano, J. C., y Medina Chicaiza, R. P. (2022). Análisis bibliométrico y
contextual de la producción científica del marketing sensorial. Religación, 7(31), e210883. https://d=
oi.org/10.46652/rgn.v7i31.883
Saravia Ramos, G. =
del
P., Palomino Flores, J. V., Mercado Hermenegildo, A., y Pichilingue Romero,=
J.
V. U. (2022). Comportamiento del consumidor: Estudio bibliométrico de la ba=
se
de datos Scopus 2011-2021. INNOVA Research Journal, 7(3), 50-74. https://doi.org/10.33890/innova.v7.n3.2022.2112=
Sarker, I. H. (2021). Machine Learning:
Algorithms, Real-World Applications and Research Directions. SN Computer
Science, 2(3), 160. https://doi.org/10.1007/s42979-021-00592-x
Sarker, I. H., Furhad=
,
M. H., y Nowrozy, R. (2021). AI-Driven Cybersec=
urity:
An Overview, Security Intelligence Modeling and Research Directions. SN Computer Scienc=
e, 2(3), 173=
. =
https://doi.org/10.1007/s42979-021-00557-0
Torres, P. (2016). Acerca =
de
los enfoques cuantitativo y cualitativo en la investigación educativa cubana
actual. Atenas, 2(34), 1-15. http://atenas.umcc.cu/index.php/atenas/article/view/194
Urbizagastegui, R.
(2021). La bibliometría brasileña: minería de textos. Revista ACB, <=
i>26(1),
1–18. =
https://revista.acbsc.org.br/racb/article/view/1768
Vasquez, C. E., y
Bolaños, J. M. (2019). Implementación y evaluación de técnicas de cluste=
ring
para un sistema de recuperación de documentos jurisprudenciales [Tesis =
de
Grado]. Fundación Universitaria De Popayán. =
https://fupvirtual.edu.co/repositorio/files/original/4fa0079ccb70d407eeac13=
6eec3a5c7ec59bae74.pdf
Wickman, H. (2016). ggplot2: Elegant
Graphics for Data Analysis. Springer Verlag. https://ggplot2.tidyver=
se.org
El artículo que se publica es de exclusiva responsabilidad de los autor=
es y
no necesariamente reflejan el pensamiento de la Revista Ciencia & Turismo.
![3Deditori=](3D"02paperok.files/image026.jpg")
El artículo queda en propiedad de la revista y, por tanto, su publicaci=
ón
parcial y/o total en otro medio tiene que ser autorizado por el director de=
la Revista Ciencia & Turismo.
![](3D"02paperok.files/image028.png")