www.ijcrsee.com
389
El Bahri, N. et al. (2023). Using students’ digital written text in Moroccan dialect for the detection of student personality factors,
International Journal of Cognitive Research in Science, Engineering and Education (IJCRSEE), 11(3), 389-400.
Introduction
Student written text plays a special and dynamic role in learning environments that uses social
media. Actually, students use social media platforms like (Instagram, WhatsApp and Twitter) more and
more, and there are many advantages of using an online learning environment for educational reasons
(Christine Greenhow et al., 2019). Those platforms enable informal and spontaneous conversation
between students. Therefore, they frequently post variety of content, and they receive in return quick
feedback and interactions with other students and teachers. Among the advantage of this trend, these
interactions encourages participation and teamwork in students, which can improve their learning process
(Josué et al., 2023). Moreover, unlike standard academic writing, students may present ideas, queries,
or observations in a less structured manner. As a result, this informal setting may foster individualism,
creativity, and the exchange of different perspectives (Eysenck, 1994).
Students’ writing in social media often reects their ideas, feelings, interests, and ways of interacting,
which can provide interesting insights about them. Thus, based on these writings, it is possible to
determine their personalities on social media using a variety of methods, including: content and interests,
language and communication style, frequency and consistency, tone and emotions, interaction patterns,
etc... (Rahman et al., 2019).
To understand individual’s personality many approaches and theories are used. The following list
presents the most well-known models:
• The Myers-Briggs Type Indicator (MBTI) (Pittenger, 1993; Tlili et al., 2016): categorizes persons
Using Students’ Digital Written Text in Moroccan Dialect For The Detection
of Student Personality Factors
Nisserine El Bahri1* , Zakaria Itahriouan2 , Anouar Abtoy1 , Samir Brahim Belhaouari3
1Abdelmalek Essadi University, Tetouan, Morocco, e-mail: nisserine.elbahri@etu.uae.ac.ma, anouar.abtoy@uae.ma
2Moulay Ismail University, Meknes, Morroco; Private University of Fez, Morroco, e-mail: itahriouan@upf.ac.ma
3Hamad Bin Khalifa University, Doha, Qatar, e-mail: sbelhaouari@hbku.edu.qa
Abstract: In the contemporary digital era, social media platforms have a big inuence on students’ lives. They use these
platforms for self-expression, opinion sharing, and experience reporting (writing or sharing videos or photos about personal
experiences) in addition to social interaction. Education professionals and academics may get valuable insights into students’
thoughts, sentiments, interests, academic success, and even personalities by studying their writing on social media. We can
improve our teaching, enhance students’ social and emotional development, and create a more engaging learning environment
if we have a better knowledge of the student. The purpose of this study is to ascertain whether or not students interact with
classmates and other participants in learning platforms in a way that accurately represents their personalities. Data from a sample
of students at Abdelmalek Essaadi University of Tetouan were collected from various social media learning environments for
the experimental investigation presented in this work, and Symanto AI-based personality tool was used to assess the data. The
Big Five Questionnaire was then utilized to assess the personalities of the same students, and the ndings were compared to
the personality traits discovered by the AI-based approach. The study has shown that the AI based tool has correctly predicted
the personality traits of 7 students out of 10 with a correlation of about 0,9 which means that social media-based learning
environments can be used by institutions to understand the personality of the student. This paper also gives recommendations
about data for obtaining good quality in personality prediction.
Keywords: FFM personalities, social media learning environment, Moroccan dialect text.
Original scientic paper
Received: September 24, 2023.
Revised: October 30, 2023.
Accepted: November 08, 2023.
UDC:
077.5-057.875:159.923.075(64)
10.23947/2334-8496-2023-11-3-389-400
© 2023 by the authors. This article is an open access article distributed under the terms and conditions of the
Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
*Corresponding author: nisserine.elbahri@etu.uae.ac.ma
www.ijcrsee.com
390
El Bahri, N. et al. (2023). Using students’ digital written text in Moroccan dialect for the detection of student personality factors,
International Journal of Cognitive Research in Science, Engineering and Education (IJCRSEE), 11(3), 389-400.
into 16 personality types based on four dichotomies which are extraversion/introversion, sensing/intuition,
thinking/feeling, and judging/perceiving. Evry combination of these caracteristics results in a distinct
personality type, such as INFP (Introverted, Intuitive, Feeling, Perceiving) or ESTJ (Extraverted, Sensing,
Thinking, Judging).
• The Big Five (BF) Personality Traits (Caprara et al., 1993; Eysenck, 1994): Also known as the
Five Factor Model (FFM). According to this Model, there are ve different dimensions of personality:
Agreeableness, Conscientiousness, Extraversion, Openness, and Neuroticism (Utami, Maharani and
Atastina, 2021). It is frequently employed in organizational and behavioral studies as well as psychology
research, offers a thorough framework for evaluating and characterizing personality traits. It is considered
that these characteristics sum up the fundamental elements of human behavior and personality.
• Freudian Personality Structure (Bronfenbrenner, 1951; Zhang, 2020): The famous Austrian
psychotherapist Sigmund Freud developed a theory of personality structure known as the Freudian
Personality Structure. According to Freud, there are three primary parts of the human mind (Id, Ego and
SuperEgo).
• Hans Eysenck’s model (Eysenck, 1991, 1981): The prominent psychologist Hans Eysenck
developed a widely known three-dimensional model of personality: Neuroticism/Emotional Stability,
Extraversion/Introversion, and Psychoticism. The eld of personality psychology has been greatly
impacted by Eysenck’s work, and his model has been widely applied in studies and personality evaluation.
• DISC (Dominance, inuence, Steadiness and Conscientiousness) personality Model (Sugerman,
2009; Utami et al., 2022): Another psychological theory for understanding and classifying human behavior
in diverse contexts is the DISC Personality Model. It categorizes people into four main personality traits,
denoted by a different letter in the acronym DISC. This model is frequently applied in work environments
and interpersonal interactions in order to foster better understanding, cooperation, and communication
among people with diverse personality types.
This work considers the Big Five model as a method of analyzing the personality of students out of
all the models previously provided since it is the most widely used model and particularly because the AI
algorithm for personality detection from text is built on it.
Through the Automated Text-Based Personality Assessment (ATBPA) (Gjurković, Vukojević and
Šnajder, 2022), articial intelligence (AI) may predict personality from text by using well-established
psychological models. These latter can determine a person’s personality traits from written content
through analyzing the writing styles, linguistic patterns, word choices, etc.… (Christian et al., 2021). The
AI models are trained using machine learning algorithms including text classication and natural language
processing. The training of the model uses mainly the annotated data from a dataset. Therefore, the
model acquires the ability to identify patterns and connections between linguistic features and personality
traits (Gjurković, Vukojević and Šnajder, 2022).
The main goal of this paper is to use one of these ATBPA tools to identify students’ personalities
based on their writing in the Moroccan dialect in social media learning environments. For this study,
we have chosen Symanto APIs as a tool. To achieve our goal, we have gathered data from students in
various social learning environments (Instagram, Twitter, WhatsApp and Google chat). This data has been
preprocessed by removing irrelevant information and then translating it into English. Subsequently, it has
been processed by the AI-based personality algorithm to predict students’ personality traits. Finally, the
students’ personality predictions obtained by the algorithm were compared to the Big Five Questionnaire
results that were gathered from the same students.
The following section presents a summary of the literature on the use of social media learning
environment. In section 3, we explain the methodology and the applied data processing approach.
Subsequently, the experiment’s ndings are presented in Part 4 followed by an analysis of the results
and a discussion in Section 5. Finally, the paper ends with a conclusion that summarizes the work and
presents the implications for further research.
Social media and education
The term “social media” refers to a modern phenomenon that includes both mobile interaction and
web-based communication with internet users via web applications (Wickramanayake and Muhammad
Jika, 2018). Thanks to how convenient it is to access these applications, the majority of people utilize
social media for a variety of purposes, including recounting experiences, communicating, and sharing
stories from their everyday lives. In the case of students, the development of Web 2.0 and the emergence
of Web 3.0 have enabled students to produce content, exchange ideas, and share knowledge. This
development is denitely igniting a revolution in the world of education (Namaziandost and Nasri, 2019).
There are now numerous social media learning environments which are frequently utilized by our
www.ijcrsee.com
391
El Bahri, N. et al. (2023). Using students’ digital written text in Moroccan dialect for the detection of student personality factors,
International Journal of Cognitive Research in Science, Engineering and Education (IJCRSEE), 11(3), 389-400.
students. Among these, the most well-liked platforms include WhatsApp, Instagram, Facebook, Wiki,
Skype, YouTube, LinkedIn, Blogs, Twitter and Google Chat,… (Swaminathan, Harish and Cherian, 2013).
These platforms, according to Lim et al., can be grouped into seven categories: media sharing, text-based,
social networking, virtual world and games, synchronous communications, conferencing applications and
mashups, and mobile-based application (See Yin Lim et al., 2014). The top 17 social media learning
environments in terms of monthly active users are listed in Table 1.
Table 1
The top 17 social media learning platforms (Barrot, 2022).
Materials and Methods
This study focuses on how interactions in learning contexts reveal students’ personalities. In the
previous section, we discussed the most common social media environments that are frequently visited
by students and how they use them to interact about learning issues. On the other hand, there is a set of
AI-based tools that can be used to detect the personality of people. These tools can be used either using
Application Programming Interfaces (APIs) or Graphical User Interfaces (GUIs). In this context, we have
gathered data from students in the classroom and we have used ‘Symanto’ (https://www.symanto.com/)
as one of these AI-powered tools to evaluate their personalities.
Detecting the personality based on the tool may not be enough to conrm that it is really the
personality of the learner. Therefore, to assess the accuracy of these latter, we asked the same students
who participated in the experiment to answer a Big Five Questionnaire test. The purpose is to compare
test results to personality traits predicted by AI. As a nal goal, we will then be able to investigate how
learning environments may be used to understand students’ personalities. The general steps of the
experiment are shown in Figure 1.
Figure 1. Summary of the student personality comparison and detection technique
www.ijcrsee.com
392
El Bahri, N. et al. (2023). Using students’ digital written text in Moroccan dialect for the detection of student personality factors,
International Journal of Cognitive Research in Science, Engineering and Education (IJCRSEE), 11(3), 389-400.
Data collection
This research involved ten students of Computer Science Engineering who were enrolled at the
National School of Applied Sciences of Tetouan at the Abdelmalek Essaadi University. We gathered their
text expressions in different contexts and from multiple social media platforms. Mainly text captured from
comments on publications (courses, labs, exercises solutions…), discussions and publications posted
by students themselves. The text was integrated and translated before being analyzed by the Machine
Learning model. Meanwhile, the same students were asked to respond to the Big Five Questionnaire test.
From numerous social media platforms, information on all target students was collected. The
targeted platforms were chosen depending on the data that is readily available for each student. For
example, concerning the rst student (see Table 2) we gathered 51 samples from Whatsapp, 41 samples
from Instagram, 6 samples from Twitter and 5 samples from Google Chat. The quantity of data samples
that were collected for each student in each social media platform is displayed in the Table 2.
Table 2
Sample of data for each student per each platform included in the study
As shown in Table 2, the number of collected data sample is not equivalent comparing different
platforms to each other’s. This is due to the fact that some platforms are more often used by students
than others (WhatsApp and Instagram for example). The total number of data gathered is 1161, with an
average of 116,1 samples for each student. The data was stored in CSV les, with each line containing
the student’s text samples arranged by their originating environment. To make sure that the content of the
students’ discussions was obvious and comprehensible and that the process of organizing the data was
completed without errors, all data was reviewed as well as we ensured the samples belonged to the right
students.
Data selection/preprocessing
In this step, we have considered data of all students taking part in the study as an adequate number
of samples was collected for each participant. In order to prevent utilizing unrepresentative samples, they
had previously been chosen using a variety of estimated characteristics.
For data cleaning, we removed some iInsignicant data which made up less than 2.8% of the
overall data. The data has been transformed before being processed due to Moroccan students’’ use of
their native dialect (“darija”) in their writings. The data transformation procedure was a very challenging
step. The text could not be used in its original form since the NLP (Natural Language Processing) Model
does not support the Moroccan dialect. To address this issue, we translated the content into English as
it is a language that the model can comprehend. Therefore, we have carried out a full English translation
of the content (see examples in Table 3). The next stage was to incorporate all the student samples,
regardless of the environment type. Subsequently, all student data were collected in one entry of the
Model because we needed to predict each student’s personality separately.
www.ijcrsee.com
393
El Bahri, N. et al. (2023). Using students’ digital written text in Moroccan dialect for the detection of student personality factors,
International Journal of Cognitive Research in Science, Engineering and Education (IJCRSEE), 11(3), 389-400.
Table 3
Text language standardization to English
Symanto: personality prediction tool
To identify and analyze personalities from the text, there are several AI-based solutions. Many of
them are a paid service for business use and can be accessed only for evaluation purpose. However,
there are some other applications that are for free use in research domain. In this work we use the free
evaluation API of Symanto which is a tool that provides companies with insights from customer data.
Symanto is also provided as an API on RapidAPI (https://rapidapi.com/).
In this experiment, we used the RapidAPI web interface to send student text, which resulted in an
HTTP POST request to the API. The web service response contains predictions made by the AI Model
presented in the .json data format. This last contains the probability of each personality trait based on the
Big Five, as well as multiple subclasses of each personality trait. Only predicted values of the rst level of
the Big Five personality traits were considered in this study.
Results
The preliminary results of this work are the outcomes originated from examining student writing
using Symanto AI-based tool to detect personality. The results include the probability of each personality
trait according to the Big Five Model which is displayed in Table 4.
Table 4
Students’ personality detected by Symanto
www.ijcrsee.com
394
El Bahri, N. et al. (2023). Using students’ digital written text in Moroccan dialect for the detection of student personality factors,
International Journal of Cognitive Research in Science, Engineering and Education (IJCRSEE), 11(3), 389-400.
The same students were also requested to complete a personality questionnaire based on the Big
Five Model, as was previously mentioned. The results of this personality test are shown in Table 5.
Table 5
Students’ personality detected based on the BF Questionnaire.
Representing personality traits detected by the AI based tool and those assumed by the
Questionnaire in a radar chart for each student can assist in comparing their results. The following gures
illustrate the radar charts buit based on the comparison between the results of the tables presented
previously (see Figure 2).
www.ijcrsee.com
395
El Bahri, N. et al. (2023). Using students’ digital written text in Moroccan dialect for the detection of student personality factors,
International Journal of Cognitive Research in Science, Engineering and Education (IJCRSEE), 11(3), 389-400.
Figure 2. Radar charts of personality predictions for each student
The above-mentioned data demonstrates that diverse personalities predicted by the AI based tool
do not always produce similar results based on the Questionnaire. Some students’ predictions (students
1, 3, 4, and 8) are quite accurate. However, there is a huge difference between predicted personality of
three students compared to their personality detected by the BF Questionnaire (Student 2, Student 5 and
Student 7).
Therefore, in order to evaluate results more precisely, we have decided to use metrics that can give
more insights from the results. Bellow we present all the metrics used:
• HPT (Highest personality traits): is when a student’s highest personality trait value from the AI-
based tool and his highest personality trait value from the Questionnaire are same, this setting is set
to true. This metric will give us a rate that describes how much the model predicts the most dominant
personality trait.
• LPT (Lowest personality traits): is when a student’s lowest personality trait value from the AI-
based tool and his lowest personality trait value from the Questionnaire are the same, this setting is set to
true. This metric will give us a rate that describes how much the model predicts the weakest personality
trait in a person.
• ME (Mean error): is a measurement of the average difference between the value of a personality
characteristic predicted by an AI-based tool and the value supplied by a Questionnaire for that same
personality trait. When this number is small, predictions are reliable and accurate.
• SD (Standard deviation): gauges the variation in the mean error of the difference between the
expected and actual values. When this score is low, there are few discrepancies in predictions for the ve
personality characteristics.
In order to obtain a more accurate result, we have assumed that unifying the precision of the
obtained results can provide a more exact metrics’ values since the precision of the values produced by
the AI-based tool (2 digits after the decimal point) and those returned by the Questionnaire (1 digit after
the decimal point) are not the same. Hence, the mean error and standard deviation were recalculated
after unifying the values.
Table 6 summarizes the results of different evaluation metrics that compares personality prediction
of Symanto AI based solution to that of the Big Five Questionnaire.
www.ijcrsee.com
396
El Bahri, N. et al. (2023). Using students’ digital written text in Moroccan dialect for the detection of student personality factors,
International Journal of Cognitive Research in Science, Engineering and Education (IJCRSEE), 11(3), 389-400.
Table 6
Evaluation metrics results of the comparison between Symanto APIs and the Big Five Questionnaire
The Symanto APIs predict correctly the highest and lowest personality traits in seven students
from a total of ten students. For all students, the mean difference is roughly 0.11, while the standard
deviation is about 0.05. The model predicts the Big Five personality characteristics well based on initial
measurements or unied precision metrics.
The objective of evaluating the correlation between predictions from the Symanto AI-based model
and those from the Questionnaire is to gure out the accuracy of Symanto APIs for each personality trait.
Therefore, instead of comparing the values themselves, the purpose is to compare the variety of the
predictions. This indicates that if the correlation is close to 1 the values predicted by the AI based tool and
those of the BF Questionnaire are very dependent. In other words, the change in the personality from
one student to another occurs in exactly the same way even if the predicted values are not completely
identical to those identied by the BF Questionnaire. This also means that for each personality trait, there
is a very strong correlation between the predictions made by the AI-based model and the outcomes of the
Questionnaire. The correlation results are shown in Table 7. It was calculated both separately for each
personality trait as well as overall for all personality traits.
Table 7
Correlation between Symanto predictions and Questionnaire results
As shown in the previous table, the correlation values of personality traits in the Symanto model
range between 0.31 and 0.84. The correlation coefcient for all traits is approximately 0.61 based on the
initial values and on the unied precision values. In general, we cannot deny the relationship between
the model’s predictions and those calculated by the Questionnaire. However, the correlation is not very
strong, balanced around 0.6.
Even the correlation between results indicated by the Questionnaire and those predicted by the AI-
based model is signicant, we cannot consider it substantial. Moreover, we had identied three students
(students 2, 5 and 7) whose ndings were considerably different when comparing Symanto AI-based
model predictions to the Questionnaire ones (see radar charts in Fig. 3). Consequently, presuming that it
www.ijcrsee.com
397
El Bahri, N. et al. (2023). Using students’ digital written text in Moroccan dialect for the detection of student personality factors,
International Journal of Cognitive Research in Science, Engineering and Education (IJCRSEE), 11(3), 389-400.
contains biased data (see discussion section), we decided to recalculate the correlation after eliminating
these three students’ data (see Table 8). The correlation increased substantially as a result of this
modication approaching 0.9.
Table 8
Correlation between Symanto predictions and Questionnaire results after elimination of biased data
We were able to assert the absurdity of the predictions made about the three students by reevaluating
the data while taking into consideration the samples which show the personality of the student. Table 9
summarizes the results of this reevaluation while table 10 shows some examples of low and good quality
samples.
Table 9
Number of samples classied by quality after the revision.
Table 10
Examples of good and low quality samples after the revision.
Discussion
Concerning the evaluation of Symanto as a personality prediction tool, we can conrm from the
beginning based on the radar charts representation that Symanto AI-based solution was successful in
predicting personality traits based on the Big Five Model. Predicted personality of seven students from
ten was very close to this obtained by the questionnaire.
www.ijcrsee.com
398
El Bahri, N. et al. (2023). Using students’ digital written text in Moroccan dialect for the detection of student personality factors,
International Journal of Cognitive Research in Science, Engineering and Education (IJCRSEE), 11(3), 389-400.
The evaluation metrics dened in this research revealed more important information about the
evaluated model. For the same seven students, the most and least dominant personality traits were
correctly predicted. The low values of the Mean Error and the Standard deviation conrm that the
predictions are very accurate.
Unifying the precision of the values obtained was supposed to give a more precise comparison
between the results. However, the ndings of the metrics recalculation following the adjustment of values
with a uniform precision did not signicantly affect the initial results. But in any case, we can consider that
this consequently conrms the initial results.
The correlation between the two compared results is a very signicant metric that calculates
numerically if the values of personality vary in the same manner in both compared approaches. A more
important insight that correlation gave us is the variation concerning each personality trait separately. The
calculation of this metric showed signicant correlation based on the results of all students for all personality
traits. Regarding the assessment of the correlation of each personality trait apart, the correlation was
raised for the Openness and the Agreeableness, on the other hand it was not very signicant for the rest
of the traits.
Although overall the results obtained clearly show that the AI-based tool can detect the personality
of students. We looked for an explanation for the failure of this process for the three students concerned.
The most likely hypothesis based on how the AI model worked was the quality of the text gathered
for these students. The text written by a person may in certain cases not reect its true personality.
This means that the personality detection algorithm may not be responsible for the error of a detected
personality from a text that does not contain expressions that really show this personality. The Symanto
team also recommends using text that specially expresses the person point of view because trough this
the person shows its personality. According to this, we decided to revise (human revision) all the data to
check the quality of the text taking in order to classify it according to this criterion. Table 9 represents the
number of good-quality and low-quality samples for each student after the revision of the data. Examples
of good and low quality text are shown in table 10. Therefore, based on the results of the new revision
we decided to exclude the data of the three students and recalculate the same evaluation metrics based
on the remaining seven students. We considered that the new results are fairer to evaluate the model
because they are not affected by low quality text.
The re-evaluation of the model based on the seven students who wrote texts that show more of their
personality yielded more accurate results. The correlation is really very strong between the personality
detected by the model and that of the questionnaire. The latter has spread 0.9 for all personality traits while
it was between 0.85 and 0.97 for separated personality traits. This conrms substantially the accuracy of
the predictions of the personality.
Generally, we can assume that this AI-based tool that employs related techniques from NLP to
identify personality from language expression predicts personality traits exceptionally well. This fact is
strongly conditioned by having sufcient data collected about each evaluated student, as well as by the
quality of the data that should include expressions that reveal the student’s thoughts and behavior.
Ultimately, as understanding students’ personalities is very important in the teaching and learning
context. The present study shows that using automated text-based personality assessment combined
with good quality text of student collected from social media-based environment can help in detecting
student personality instead of using the traditional Big Five Questionnaire. and therefore, thanks to these
tools we can develop specic solutions that automatically detect the learner’s personality in social media-
based learning environments.
Conclusions
In the current study, we collected data from social networks where students discuss about learning.
We specically gathered data from Twitter, Google Chat, Instagram, and WhatsApp groups. Data was
processed using the AI-based Symanto APIs in order to identify students’ personalities. The outcomes of
this latter were analyzed to determine if the combination of student data from learning environments and
this tool can help in identifying student’s personality. The results from the Symanto AI-based tool were
compared to results from the personality test Questionnaire.
According to the study’s outcomes, Symanto APIs can be really helpful in determining a student’s
personality based on their interactions in learning environments. Most of the student’s personality traits
were implied in their written text. Moreover, text quality is the most important factor in determining a
student’s personality.
www.ijcrsee.com
399
El Bahri, N. et al. (2023). Using students’ digital written text in Moroccan dialect for the detection of student personality factors,
International Journal of Cognitive Research in Science, Engineering and Education (IJCRSEE), 11(3), 389-400.
The evaluated tool accurately identied the dominant personality characteristic and almost every
other trait with a very low error value. Furthermore, the correlation between projected personality from text
and those found by the Questionnaire was very signicant (around 0,9) after omitting three students who
were considered as outlier samples. Either for personality in general or for individual personality traits, the
correlation value is considerably strong.
This experiment also shows that not all of the students’ personalities can accurately be identied
from text. Out of 10 students, three had personalities that the AI-based tool was unable to identify. This
failure can be explained by the text itself more often than by the tool’s inefciency. The data of these
students does not contain sufcient expressions that show explicitly students’ personality traits. Therefore,
it is strongly recommended to check the quality of the text before processing it by the model to detect the
personality.
In future work, we will evaluate other AI personality detection tools and test them on other data to
be collected. We will also study other personality detection techniques that are not text-based to ultimately
build a multimodal solution that detects personality from several data sources.
Acknowledgements
This work was supported by “Centre National pour la Recherche Scientique et Technique” (Grant
agreement number SHSE-2021/49).
Conict of interests
The authors declare that they have no known competing nancial interests or personal relationships
that could have appeared to inuence the work reported in this paper.
Author Contributions
Data Curation, N. El Bahri, A. Abtoy; methodology, Z. Itahriouan, N. El Bahri; Formal analysis,
N. El Bahri, Z. Itahriouan, S. Brahim Belhaouari; Validation, Z. Itahriouan, A. Abtoy; writing—original
draft preparation, N. El Bahri; writing—review and editing, Z. Itahriouan, A. Abtoy, S. Brahim Belhaouari;
Funding acquisition, Z. Itahriouan. All authors have read and agreed to the published version of the
manuscript.
References
Barrot, J. S. (2022). Social media as a language learning environment: a systematic review of the literature (2008-2019).
Computer assisted language learning, 35(9), 2534-2562. https://doi.org/10.1080/09588221.2021.188673
Bronfenbrenner, U., (1951). Toward an integrated theory of personality, in: Perception: An Approach to Personality. Ronald
Press Company, New York, NY, US, pp. 206–257. https://doi.org/10.1037/11505-008
Caprara, G. V., Barbaranelli, C., Borgogni, L., & Perugini, M. (1993). The “Big Five Questionnaire”: A new questionnaire to
assess the ve factor model. Personality and individual Differences, 15(3), 281-288. https://doi.org/10.1016/0191-
8869(93)90218-R
Christian, H., Suhartono, D., Chowanda, A., Zamli, K.Z. (2021). Text based personality prediction from multiple social media
data sources using pre-trained language model and model averaging. Journal of Big Data 8, 68. https://doi.org/10.1186/
s40537-021-00459-1
Christine Greenhow, Sarah M. Galvin, K. Bret Staudt Willet, (2019). What Should Be the Role of Social Media in Education?
[WWW Document]. Retrieved from https://journals.sagepub.com/doi/abs/10.1177/2372732219865290 (accessed
9.5.2023).
Digman, J. M. (1990). Personality structure: Emergence of the ve-factor model. Annual review of psychology, 41(1), 417-440.
https://doi.org/10.1146/annurev.ps.41.020190.002221
Hans, C., Suhartono, D., Andry, C., & Zamli, K. Z. (2021). Text based personality prediction from multiple social media data
sources using pre-trained language model and model averaging. Journal of Big Data, 8(68). https://doi.org/10.1186/
s40537-021-00459-1
Eysenck, H. J. (1981). General features of the model. In A model for personality (pp. 1-37). Berlin, Heidelberg: Springer Berlin
Heidelberg. https://doi.org/10.1007/978-3-642-67783-0
Eysenck, H. J. (1991). Dimensions of personality: The biosocial approach to personality. In Explorations in temperament:
International perspectives on theory and measurement (pp. 87-103). Boston, MA: Springer US. https://doi.
org/10.1007/978-1-4899-0643-4_7
Eysenck, H. J. (1994). The “Big Five” or “Giant 3”? Criteria for a paradigm. The developing structure of temperament and
personality from infancy to adulthood. Hillsdale, NJ: Erlbaum.
Gjurković, M., Vukojević, I., & Šnajder, J. (2022). SIMPA: statement-to-item matching personality assessment from text. Future
Generation Computer Systems, 130, 114-127. https://doi.org/10.1016/j.future.2021.12.014
www.ijcrsee.com
400
El Bahri, N. et al. (2023). Using students’ digital written text in Moroccan dialect for the detection of student personality factors,
International Journal of Cognitive Research in Science, Engineering and Education (IJCRSEE), 11(3), 389-400.
Greenhow, C., Galvin, S. M., & Staudt Willet, K. B. (2019). What should be the role of social media in education?. Policy
Insights from the Behavioral and Brain Sciences, 6(2), 178-185. https://doi.org/10.1177/2372732219865290
Josué, A., Bedoya-Flores, M. C., Mosquera-Quiñonez, E. F., Mesías-Simisterra, Á. E., & Bautista-Sánchez, J. V. (2023).
Educational Platforms: Digital Tools for the teaching-learning process in Education. Ibero-American Journal of
Education & Society Research, 3(1), 259-263. https://doi.org/10.56183/iberoeds.v3i1.626
Namaziandost, E., & Nasri, M. (2019). The impact of social media on EFL learners’ speaking skill: a survey study involving EFL
teachers and students. Journal of Applied Linguistics and Language Research, 6(3), 199-215. https://www.jallr.com/
index.php/JALLR/article/view/1031
Pittenger, D. J. (1993). Measuring the MBTI… and coming up short. Journal of Career Planning and Employment, 54(1), 48-52.
https://img3.reoveme.com/m/614576efb2b91676.pdf
Rahman, M. A., Al Faisal, A., Khanam, T., Amjad, M., & Siddik, M. S. (2019, May). Personality detection from text using
convolutional neural network. In 2019 1st international conference on advances in science, engineering and robotics
technology (ICASERT) (pp. 1-6). IEEE. https://doi.org/10.1109/ICASERT.2019.8934548
See Yin Lim, J., Agostinho, S., Harper, B., & Chicharo, J. (2014). The engagement of social media technologies by undergraduate
informatics students for academic purpose in Malaysia. Journal of Information, Communication and Ethics in Society,
12(3), 177-194. https://doi.org/10.1108/JICES-03-2014-0016
Sugerman, J. (2009). Using the DiSC® model to improve communication effectiveness. Industrial and Commercial Training,
41(3), 151-154. https://doi.org/10.1108/00197850910950952
Swaminathan, T. N., Harish, A., & Cherian, B. (2013). Effect of social media outreach engagement in institutions of
higher learning in India. Asia-Pacic Journal of Management Research and Innovation, 9(4), 349-357. https://doi.
org/10.1177/2319510X14523101
Tlili, A., Essalmi, F., Jemni, M., & Chen, N. S. (2016). Role of personality in computer based learning. Computers in Human
Behavior, 64, 805-813. https://doi.org/10.1016/j.chb.2016.07.043
Utami, E., Hartanto, A. D., Adi, S., Oyong, I., & Raharjo, S. (2022). Proling analysis of DISC personality traits based on Twitter
posts in Bahasa Indonesia. Journal of King Saud University-Computer and Information Sciences, 34(2), 264-269.
https://doi.org/10.1016/j.jksuci.2019.10.008
Utami, N. A., Maharani, W., & Atastina, I. (2021). Personality classication of facebook users according to big ve personality
using SVM (support vector machine) method. Procedia Computer Science, 179, 177-184. https://doi.org/10.1016/j.
procs.2020.12.023
Wickramanayake, L., & Muhammad Jika, S. (2018). Social media use by undergraduate students of education in Nigeria: A
survey. The Electronic Library, 36(1), 21-37. https://doi.org/10.1108/EL-01-2017-0023
Zhang, S. (2020, April). Psychoanalysis: The inuence of Freud’s theory in personality psychology. In International Conference
on Mental Health and Humanities Education (ICMHHE 2020) (pp. 229-232). Atlantis Press. https://doi.org/10.2991/
assehr.k.200425.051
