Article

The Use of a Large Language Model for Cyberbullying Detection

Bayode Ogunleye 1,* and Babitha Dharmaraj 2

1 Department of Computing & Mathematics, University of Brighton, Brighton BN2 4GJ, UK

2 Department of Digital, Data and Technology, Ofgem, London E14 4PU, UK
* Correspondence: [email protected]

Abstract: The dominance of social media has added to the channels of bullying for perpetrators.
Unfortunately, cyberbullying (CB) is the most prevalent phenomenon in today’s cyber world, and
is a severe threat to the mental and physical health of citizens. This opens the need to develop a
robust system to prevent bullying content from online forums, blogs, and social media platforms to
manage the impact in our society. Several machine learning (ML) algorithms have been proposed
for this purpose. However, their performances are not consistent due to high class imbalance
and generalisation issues. In recent years, large language models (LLMs) like BERT and RoBERTa
have achieved state-of-the-art (SOTA) results in several natural language processing (NLP) tasks.
Unfortunately, the LLMs have not been applied extensively for CB detection. In our paper, we
explored the use of these models for cyberbullying (CB) detection. We have prepared a new dataset
(D2) from existing studies (Formspring and Twitter). Our experimental results for dataset D1 and D2
showed that RoBERTa outperformed other models.

Keywords: BERT; cyberbullying; large language model; machine learning; natural language process-
ing; online abuse; RoBERTa; social media analytics

1. Introduction
The emergence of social technologies like Facebook, Twitter, TikTok, WhatsApp,
Threads, and Instagram has improved communication amongst the people and businesses
across the globe. However, despite the huge advantage of these platforms, they have also
Citation: Ogunleye, B.; Dharmaraj, B. added channels of bullying for perpetrators. Cyberbullying (CB), often referred to as online
The Use of a Large Language Model bullying, is becoming an important issue that requires urgent attention. For illustration, in
for Cyberbullying Detection. the USA, the Pew Research Center [1] reported that around two-thirds of US adolescents
Analytics 2023, 2, 694–707. https:// have been subjected to cyberbullying. Statista [2] reported in their survey that 41% of
doi.org/10.3390/analytics2030038 adults in the USA had experienced cyberbullying. The Pew Research Center [3] reported
Academic Editor: Domenico Ursino that 46% of teens in the USA aged 13 to 17 have been cyberbullied. The Office for National
Statistics [4] reported that 19% of children aged 10 to 15 (this equates to 764,000 children)
Received: 15 June 2023 have experienced cyberbullying in England and Wales. Patchin and Hinduja [5] found
Revised: 13 July 2023
out that 90% of tweens (9 to 12 years old) utilise social media or gaming apps, and 20% of
Accepted: 31 August 2023
tweens are involved in CB either as a victim, an offender, or a bystander.
Published: 6 September 2023
This problem of cyberbullying (CB) is a relatively new trend that has recently gained
more popularity as a subject. Cyberbullying is repetitive, aggressive, targeted, and in-
tentional behaviour aimed at hurting an individual’s or a group’s feelings through an
Copyright: © 2023 by the authors. electronic medium [6,7]. CB takes many forms, including flaming, harassment, denigration,
Licensee MDPI, Basel, Switzerland. impersonation, exclusion, cyberstalking, grooming, outing, and trickery [6,8]. Cyberbul-
This article is an open access article lies are more likely to be technologically astute than physically stronger, making them
distributed under the terms and better able to access victims online, conceal their digital footprints, and become involved
conditions of the Creative Commons in posting rumours, insults, sexual comments, threats, a victim’s private information, or
Attribution (CC BY) license (https:// derogatory labels [9]. The fundamental causes of any bullying incident are the imbalance of
creativecommons.org/licenses/by/ power and the victim’s perceived differences in race, sexual orientation, gender, socioeco-
4.0/). nomic level, physical appearance, and mannerism. Xu et al. [10] stated that CB participants

Analytics 2023, 2, 694–707. https://doi.org/10.3390/analytics2030038 https://www.mdpi.com/journal/analytics

Analytics 2023, 2 695

could play the role of either a bully, victim, bystander, bully assistant, reinforcer, reporter,
or accuser. Prior studies found out that CB impacts anxiety [11–13], depression [14,15],
social isolation [16], suicidal thoughts [17–19], and self-harm [20,21]. Messias et al. [22]
stated victims of cyberbullying have higher rates of depressive illnesses and suicidality
than victims of traditional bullying. Patchin and Hinduja [5] stated CB victims admit that
they frequently feel awkward or afraid to attend school, and it impacts their academic per-
formance. In addition, they found out that nearly 70% of teens who reported being victims
of cyberbullying stated it had a negative impact on their self-esteem, and nearly one-third
claimed it had an impact on their friendships. Despite the impact and increasing rate of CB,
unfortunately, there is limited attention paid to developing sophisticated approaches for
automatic CB detection.
CB studies are yet to extensively explore the use of large language models (LLMs) for
CB detection [23]. CB is commonly misinterpreted, leading to flawed systems with little
practical use. Additionally, several studies only evaluated using swear words to filter CB,
which is only one aspect of this topic, and swear words may not always indicate bullying
on platforms with a high concentration of youngsters [6,24]. Thus, it is practically useful for
developers and media handlers to have a robust system that understand context better, to
enhance CB detection. In our study, we aim to evaluate the performance of large language
models for CB detection. Unfortunately, there are some obstacles to CB detection. One is
the issue of unavailable balanced and enriched benchmark datasets [6,23,25]. The issue of
class imbalance has been a popular problem in machine learning (ML) applications, as the
ML algorithms tend to be biased towards the majority class [26]. Past studies emphasised
on the class imbalance problem in the CB context [27]. In most studies, the proportion
of bullying posts is in the range of 4–20% of the entire dataset compared to non-bullying
posts [6,28–30]. This opens the need to create a new, enriched dataset with balanced classes
for effective CB detection and make it publicly available. To this end, we propose the use
of Robustly optimized BERT approach (RoBERTa), a pre-trained large language model
for cyberbullying detection. Thus, our contributions can be summarised as follows. We
prepared a new dataset (D2) from existing studies for the development of algorithms on CB
detection. We conducted an experimental comparison of sophisticated machine learning
algorithms with two datasets (D1 and D2). We ascertained RoBERTa as the state-of-the-art
(SOTA) method for automated cyberbullying detection. The rest of the paper is organised
as follows. Section 2 will review the literature to provide background knowledge to this
study. Section 3 will present the methodology. Section 4 will present and discuss the results,
and Section 5 will provide conclusions and recommendations.

2. Related Work
Cyberbullying (CB) is the most prevalent phenomenon in today’s digital world, and is
a severe threat to the mental and physical health of cybercitizens [14,31]. Several studies
have proposed various techniques for automated CB detection. For example, the authors
in [10] crawled 1762 tweets from Twitter using keywords such as “bully, bullied, bullying”.
The data was labelled by five human annotators such that 684 were labelled as bullying and
1078 as non-bullying. They compared four traditional machine learning models, namely
Naïve Bayes (NB), Support Vector Machines (linear SVM), Support Vector Machines (RBF—
SVM), and Logistic regression (LR). Their result showed linear SVM achieved the best
performance with a 77% F1 score. Agrawal and Awekar [28] compared machine learning
(ML) models, namely Naïve Bayes (NB), Support Vector Machines (SVM), random forest
(RF), convolutional neural network (CNN), long short-term memory (LSTM), bidirectional
long short-term memory (BiLSTM), and BiLSTM with an attention mechanism. They
used datasets from three different social media platforms namely, Formspring (a total of
12,773, split into 776 bully and 11,997 non-bullying text), which was collected by the authors
in [8], Twitter (16,090, split into 1937 bullying through racism, 3117 bullying through sexism,
and 11,036 non-bullying text), collected by the authors in [32], and Wikipedia (115,864, split
into 13,590 attack and 102,274 non-attack text), collected by authors in [33]. They oversam-
Analytics 2023, 2 696

pled the minority class using the Synthetic Minority Oversampling technique (SMOTE).
Their BiLSTM with attention implementation achieved an F1 score of at least 87% for the
bullying across the three different social media platforms. Similarly, the authors in [34] re-
produced the experiment in Agrawal and Awekar [28] with the Formspring dataset [8] only.
Their results showed that SVM performed better than logistic regression (LR), decision tree
(DT), and random forest (RF), with 98% accuracy and an F1 score of 93% (86% F1 score for
bullying class). Alduailaj and Belghith [35] compared SVM and NB for cyberbullying detec-
tion in an Arabic language context. They collected 30,000 Arabic comments on 5 February
2021 from Twitter and YouTube, and the comments were labelled manually as bullying and
non-bullying using most common and frequent Arabic bullying keywords detected from
the comments. Their result showed term frequency inverse document frequency (TF-IDF)
vectors deployed to SVM achieved accuracy of 95% and an F1 score of 88%. Lepe-Faúndez
et al. [36] proposed a hybrid approach for CB (aggressive text) detection in a Spanish lan-
guage context. They compared twenty-two hybrid models, a combination of lexicons and
machine learning algorithms, across three datasets, namely Chilean (1013 aggressive and
1457 non-aggressive tweets), Mexican (2112 aggressive and 5220 non-aggressive tweets),
and Chilean–Mexican (3127 aggressive and 6675 non-aggressive tweets) corpora. In their
experiment, they tested the approaches with 30% of the dataset and showed a hybrid
approach with lexicon features achieved superior performance, and models with SVM as
classifier also achieved better performance amongst the ML algorithms deployed. Dewani
et al. [37] showed that SVM and embedded hybrid N-gram approach performed best in
detecting cyberbullying in the Roman Urdu language context, with an accuracy of 83%.
Suhas-Bharadwaj et al. [38] applied extreme learning machine to classify cyberbullying
messages, and achieved accuracy of 99% and an F1 score of 91%. Woo et al. [39] conducted
a systematic review of CB literature. Their literature review findings suggest SVM and
Naïve Bayes (NB) are the best-performing models for CB detection.
Recently, there has been development of large language models (LLMs), which have
taken the world by surprise. The LLMs have been applied to several NLP tasks like topic
modelling [40], sentiment analysis [41], a recommendation system [42], and harmful news
detection [43]. In the context of CB detection, Paul and Saha [44] compared bidirectional
encoder representations from transformers (BERT) to BiLSTM, SVM, LR, CNN, and a hybrid
of RNN and LSTM, using three real-life CB datasets. The datasets are from Formspring [8],
Twitter [32], and Wikipedia [33]. They used SMOTE to rebalance the dataset and, thus,
showed BERT outperformed other models across the datasets with an F1 score of at least
91%. Similarly, Yadav et al. [45] applied BERT to the Formspring dataset [8] and Wikipedia
dataset [33]. Their approach achieved an F1 score of 81% for the Wikipedia dataset. They
rebalanced the Formspring dataset thrice, and achieved an F1 score of 59% with the
first oversampling rate, an F1 score of 86% with the second oversampling dataset, and
a 94% F1 score in the third oversampling dataset. However, it is worth mentioning that
they have tested their model on the oversampled dataset, and thus might not be reliable in
terms of generalisation. Yi and Zubiaga [41] used the same dataset from Formspring [8],
Wikipedia [33], and Twitter [32]. They proposed XP-CB, a novel cross-platform adversarial
framework based on transformers and adversarial learning models for cross-platform CB
detection. They showed that XP-CB could enhance a transformer leveraging unlabelled
data from the source and target platforms to come up with a common representation
while preventing platform-specific training. They showed XP-CB achieved an average
macro F1 score of 69%. In summary, popular data sources for CB detection are Wikipedia,
Formspring, and Twitter [8,27,44,45]. Our literature review findings suggest that very few
studies have used the transformers models (pre-trained large language models) for CB
detection. A CB literature survey conducted by Woo et al. [39] found that most studies
have used traditional machine learning models for CB detection. Thus, our paper looks to
compare the performance of SOTA language models for CB detection.
Analytics 2023, 2 697

3. Methodology
We propose the use of the fine-tuned Robustly optimized BERT approach (RoBERTa)
for automatic cyberbullying (CB) detection. We conducted an experimental comparison
of large language models to traditional machine learning models, such as support vector
machine (SVM) and random forest (RF).

3.1. Bidirectional Encoder Representations from Transformers (BERT)

BERT [46] is a self-supervised autoencoder (AE) language model developed by Google
in 2018 for training NLP systems. BERT is a bidirectional transformer-based model pre-
trained on a large-scale Wikipedia corpus using the Masked Language Model (MLM) and
next-sentence prediction tasks. BERT (base) is implemented based on transformer and
attention mechanism, which uses encoders to read in input and decoders to output. The
BERT base model consists of 12 layers of transformer blocks, a hidden layer size of 768, and
12 self-attention heads. The model comprises of two stages, which are the pre-training stage
and the fine-tuning stage. Without making significant task-specific architecture alterations,
the pre-trained BERT model can be improved with just one extra output layer to produce
cutting-edge models for a variety of tasks, including question answering and language
inference. The objective of the masked language model is to predict the actual vocabulary
identity of a masked word based only on its context after randomly masking some of the
tokens from the input. The MLM’s intent permits the representation to combine the left
and the right context, in contrast to the left-to-right language model pre-training, which
allows us to pre-train a deep bidirectional Transformer. To fine-tune the pre-trained BERT
model, the model is first instantiated with default parameters (used when pre-trained),
and then the parameters are fine-tuned using labelled data from downstream tasks (text
classification in our case). Every sequence will start with a particular classification token as
the first token ([CLS]). For classification tasks, the last hidden state matching to this token
is used as the aggregate sequence representation (Figure 1 below). The sum of the token
embeddings, segmentation embeddings, and position embeddings constitutes the input
embeddings, as shown in Figure 2 below.

Figure 1. An illustration of BERT model architecture.

Analytics 2023, 2 698

Figure 2. BERT input representation [46].

3.2. RoBERTa
Facebook AI Research (FAIR) identified the limitations of Google’s BERT and proposed
the Robustly optimized BERT approach (RoBERTa) in 2019. Liu et al. [47] stated that BERT
was undertrained, and they modified the training method by (i) using a dynamic masking
pattern instead of static, (ii) training with more data with large batches, (iii) removing next
sentence prediction, and (iv) training on longer sentences, and proposed RoBERTa. As a
result, RoBERTa outperforms BERT in terms of the masked language modelling objective
and performs better on downstream tasks. For training the model, the researchers employed
pre-existing unannotated NLP datasets, as well as the unique collection CC-News, which
was compiled from publicly available news stories. RoBERTa is a component of the effort
by Facebook to improve the state-of-the-art in self-supervised models so that they may be
created with less dependency on time and data-labelling.

3.3. XLNet
XLNet [48] is a permutation based autoregressive transformer that combines the
finest aspects of autoencoding and autoregressive language modelling while seeking to get
around their drawbacks. BERT (as an autoencoder language model) ignores the dependency
between the masked positions and leads to a pretrain–finetune discrepancy because it relies
on masking the input to corrupt it. On the other hand, conventional autoregressive (AR)
language models predict the next word based on the word’s context, either in forward or
backward direction, but not in both. XLNet’s training objective calculates the likelihood of
a word based on all possible word permutations in a sentence, rather than only those to the
left or right of the target token. Integrating Transformer-XL and a carefully thought-out two-
stream attention mechanism are just two of the ways that the XLNet neural architecture
is designed to perform in perfect harmony with the autoregressive (AR) mission. It is
anticipated that, to capture bidirectional context, each position would learn to make use of
contextual data from all positions.

3.4. XLM-RoBERTa
The multilingual version of RoBERTa is called XLM-RoBERTa [49], which was released
by Facebook as an update to their XLM-100 model. It was trained on 100 languages from
2.5 TB of filtered common crawl data. The “RoBERTa” part in XLM-RoBERTa originates
from the fact that it uses the identical training procedures as the monolingual RoBERTa
model, with the Masked Language Model training objective. There is no ALBERT-style
sentence order prediction or BERT-style Next Sentence prediction in XLM-RoBERTa.
Analytics 2023, 2 699

3.5. Support Vector Machine

In the 1990s, Vapnik [50] proposed the Support Vector Machine (SVM). The super-
vised learning algorithm has been shown to be a reliable and well-performing model for
classification tasks [26,51]. An SVM constructs a hyperplane between separated marginal
lines of the nearest support vectors (input vector) of the classes. In the space, an optimal
separating hyperplane is determined by maximizing the marginal distance. An SVM is
a linear classifier that also performs a non-linear classification problem using the kernel
function. In the non-linear classification problem, the algorithm maps the data into input
vector and uses the kernel function to transform low dimensional input space to a higher
dimension to solve non-linear problems. The maximum margin hyperplane is used for op-
timal classification of a new instance (class). The higher the marginal distance between the
classes, the more generalizable the result is. Figure 3 below provides a visual representation
of an SVM.

Figure 3. An illustration of a Support Vector Machine [51].

3.6. Random Forest

Breiman [52] introduced random forest (RF). RF is a robust machine learning algo-
rithm that uses an ensemble learning approach. RF is an ensemble method that uses a
random subset of features (from a training set) to train multiple independent decision trees
(bootstrap) and predict instance by majority voting of each tree outcome or the average.
The model is easy to interpret, runs efficiently on large database, is fast to train and scalable,
performs well in complex datasets, and is robust to irrelevant features [53–55]. However,
it is sensitive to overfitting, which can then be regulated using the numbers of trees. RF
can be used for both classification and regression problems. Furthermore, the model can
accommodate both categorical and continuous data and, most importantly, handles miss-
ing values and outliers well. Figure 4 below shows an illustration of the algorithm. The
algorithm’s iteration can be summarized as follows:
Analytics 2023, 2 700

• Select k data points randomly;

• Build decision trees with the k data points;
• Predict y value based on aggregate of class (outcome);
• Voting or average.

Figure 4. An illustration of random forest classifier [53].

3.7. Evaluation Metrics

This section discusses the evaluation metrics of the models to understand how well
they have performed in this context and help decide on the best model. In a classification
task, the common evaluation metrics are accuracy, precision, recall, and F1 measure. The
proportion of correctly and wrongly labelled classes will be described in the form of true
positive, true negative, false positive, and false negative.
Where:
True positive (TP): positive class classified correctly;
True negative (TN): negative class classified correctly;
False positive (FP): negative class wrongly predicted as positive (Type I error);
False negative (FN): positive class wrongly predicted as negative (Type II error);
Accuracy: the ratio of the number of samples predicted correctly to the total number
of samples:
TP + TN
(1)
TP + TN + FP + FN
Precision: denoted as P is the percentage % of selected items that is correct.

TP
(2)
TP + FP
Analytics 2023, 2 701

Recall: denoted as R is the percentage % of correct items that are selected.

TP
(3)
TP + FN
The F1 measure provides the balance between the precision and recall and can be
denoted as
2PR
(4)
P+R
3.8. Dataset
In this study, we have used data from existing cyberbullying (CB) studies. The datasets
were collected from Formspring.me and Twitter. We have named the datasets D1 and D2,
for easy identification. In Dataset D1, we used the dataset of Agrawal and Awekar [28].
They collected the data from Formspring and employed three human annotators from the
Amazon Mechanical Turk service to label the data as bullying or non-bullying. The data is
publicly available via this link: https://github.com/sweta20/Detecting-Cyberbullying-
Across-SMPs (accessed 24 April 2023). Table 1 below shows the class distribution of the
dataset (D1).

Table 1. D1—Formspring (CB) data distribution [28].

Label Count
Not Cyberbullying (CB) 11,997
Cyberbullying 776

Wang et al. [56] prepared a new dataset from six existing studies [10,28,32,57–59] Their
datasets were manually annotated into fine-grained CB target classes, namely, victim’s
age, ethnicity, gender, religion, other quality of CB, and not cyberbullying (notcb). The
datasets were imbalanced; hence, they applied a modified Dynamic Query Expansion
(DQE) to augment the dataset in a semi-supervised manner. They then randomly sampled
approximately 8000 tweets from each class to have a balanced dataset of approximately
48,000 tweets. The data set is publicly available, and can be accessed via the link below.
Table 2 below provides the class distribution of their dataset https://drive.google.com/
drive/folders/1oB2fan6GVGG83Eog66Ad4wK2ZoOjwu3F (accessed 24 April 2023).

Table 2. Twitter (CB) data distribution [56].

Label Count
Age 7992
Ethnicity 7961
Gender 7973
Religion 7998
Other CB 7823
Not CB 7945

The issue of data scarcity and class imbalance is a popular problem in the CB detection
domain. To resolve this, we took a different approach to Wang et al. [56]. This is because
we aim to prepare a dataset that is comparable to prior and future work, and to enhance
the development of CB detection algorithms. Thus, we prepared our binary classification
dataset D2 from existing CB studies, including Wang et al. [56]. In dataset D2, we con-
verted the multi-class dataset of Wang et al. [56] into binary classes by labelling the ‘age’,
‘ethnicity’, ‘gender’, ‘religion’, and ‘other cyberbullying’ classes as “bullying”. Agrawal
and Awekar [28] have 11,997 CB non-bullying instances, and only 776 bullying instances
from FormSpring. Thus, we concatenated the two imbalanced binary class datasets to
create ours. The instances that had less than three words or more than 100 words have been
Analytics 2023, 2 702

considered outliers and removed to obtain a more representational dataset. Table 3 below
presents the distribution of our dataset.

Table 3. D2—distribution (Our study).

Label Count Source Annotation

Bullying 19,553 FormSpring + Twitter Manual
Non-Bullying 19,526 FormSpring + Twitter Manual

3.9. Experimental Setup

The term frequency inverse document frequency (TF-IDF) feature vectors and Sentence
BERT (SBERT) embeddings [60] are used as input to the traditional machine learning mod-
els. The TF-IDF was implemented using both unigram and bigram (ngram_range = 1, 2),
the min_df is 5, and the max_df set to 0.8. The SBERT model used was ‘all-MiniLM-L6-v2′.
We deployed random forest and support vector machine algorithms using scikit-learn
package. The latter used ‘radial basis function’ (RBF) as the kernel function with C = 5 and
gamma = 0.5. The random forest classifier was employed with the number of estimators as
100, criterion is ‘gini’, and the bootstrap parameter set to ‘True’ (to allow taking subsamples
for each decision tree). We employed GridSearchCV package for the parameter tuning for
optimal solution. Additionally, we utilized HuggingFace transformers to implement pre-
trained language models. The number of training epochs used is four. We have provided
links to our source code (in the statement section) to evidence details of the transformer
architecture implemented. For training and testing, we divided the data into stratified
samples of 90% and 10%. With all the above descriptions for the models, the study was
conducted in two cases. In case 1, we used dataset D1, which is highly imbalanced with
10,702 negative class instances and 703 positive class instances for training. This imbalanced
dataset was given as input to the pre-trained language models, namely BERT, RoBERTa,
XLNET, and XLM-RoBERTa. Similarly, in case 2, the dataset D2 was used as input to the
ML algorithms, and we present the result in the subsequent section.

4. Experimental Results
This section presents the results of our experimental comparison of the classification
algorithms. Our experiment was twofold. In the first experiment, we implemented the
algorithms with the imbalanced dataset (D1), and named this case 1. In the second experi-
ment, we implemented the algorithms with the balanced dataset (D2) prepared, and this is
named case 2. Table 4 below presents the evaluation report of our case 1.

Table 4. Evaluation report of classification algorithms (case 1).

Macro
Algorithms Class Training Size Accuracy (%) Precision (%) Recall (%)
F1-Score (%)
TF-IDF + RF 0 10,792 0.80 0.89 0.75 0.82
1 703 0.41 0.27 0.34
TF-IDF + SVM 0 10,792 0.84 0.87 0.85 0.86
1 703 0.49 0.43 0.46
BERT 0 10,792 0.96 0.98 0.98 0.98
1 703 0.66 0.63 0.64
XLNet 0 10,792 0.95 0.97 0.99 0.98
1 703 0.70 0.41 0.52
RoBERTa 0 10,792 0.95 0.98 0.98 0.98
1 703 0.63 0.68 0.66
XLM - 0 10,792 0.94 0.98 0.96 0.97
RoBERTa 1 703 0.53 0.68 0.60
Analytics 2023, 2 703

In our experiment, the positive class instances (‘bully’) are denoted as ‘1 ′, and this
class is of interest. Firstly, we implemented the traditional machine learning classifiers,
namely support vector machine (SVM) and random forest (RF). The results in Table 1 above
shows that RoBERTa achieved the best performance, with an F1 score of 0.66. Other LLMs
achieved comparable results, except XLNet.
The traditional machine learning models performed poorly. We performed hyper-
parameter tuning of the models for optimal performance. Unfortunately, there was no
significant improvement. The optimal models achieved varied F1 scores at a range of
±0.064 to the published results in Table 1. In general, the algorithms struggled with the
positive class instances (‘bully’) when compared to negative instances (‘non bully’), espe-
cially the XLNet model. This is unsurprising, as its due to the class imbalance of the dataset
(D1). However, RoBERTa showed a better result. The performance might be due to the
improvement of RoBERTa on language understanding as the model is trained on larger text,
bigger batch sizes, and longer sequence compared to BERT. Our results are superior to the
results in previous studies. Agrawal and Awekar [28] applied bidirectional long short-term
memory (BiLSTM) with an attention mechanism, and their implementation achieved an
F1 score of 0.51 for the positive class instance (‘bully’). In comparison to our study, all our
large language models (LLMs) applied to D1 showed better performances.
Furthermore, in the experiments of Agrawal and Awekar [28], they improved on D1
by oversampling the minority class using the Synthetic Minority Oversampling technique
(SMOTE). Their BiLSTM implementation achieved an F1 score of 0.91 for the positive class
instance (‘bully’). However, Emmery et al. [6] criticised their implementation. Emmery
et al. [6] reproduced their experiment and discovered the overlap between train–test data
due to the oversampling method. They showed the results of Agrawal and Awekar [28] are
not reliable for the oversampled case. Thus, Emmery et al. [6] modified the implementation
(BiLSTM with an attention mechanism) by oversampling only the training data, and
achieved F1 score of 0.33 for the ‘bully’ class. To conclude for case 1 (D1), the LLMs,
namely BERT, RoBERTa, XLNet, and XLM-RoBERTa, achieved better performance than
that of Agrawal and Awekar [28] and Emmery et al. [6], as reported in Table 4 above. The
performance of the LLMs can be attributed to their power to understand context better
and effective understanding of long sequence text. Thus, it is not surprising to see that
the models have performed better than the traditional machine learning models and the
hybrid algorithms.
Table 5 below presents case 2 of our experiment. Using our dataset (D2), we ascertain
RoBERTa as the state-of-the-art model, as the algorithm achieved the best performance
overall, with an F1 score of 0.87 for the positive class (‘bully’). Also, the results of all
four language models prove that they are all comparable in performance with not much
variance. This agrees with the result of Paul and Saha [44] that showed Bidirectional
Encoder Representation from Transformer (BERT) performs better than deep learning
algorithms like BiLSTM for automated cyberbullying detection. Our experimental findings
showed that pre-trained language models are powerful and competitive with other models
in the detection of cyberbullying on social media sites. Furthermore, it is worth noting that
because we have used a balanced training dataset (D2), the traditional machine learning
models also showed good performance; most notably, the support vector machine (SVM)
achieved an F1 score of 0.85 for the positive class (‘bully’). This is consistent with the results
of Ogunleye [26] which showed SVM is a robust classification algorithm when fed with a
balanced training dataset. In summary, we propose the use of RoBERTa for CB detection
due to its consistent performance in both experiments (cases 1 and 2).
Analytics 2023, 2 704

Table 5. Evaluation report of classification algorithms (case 2).

Macro
Algorithms Class Training Size Accuracy (%) Precision (%) Recall (%)
F1-Score (%)
BERT 0 17,573 0.85 0.85 0.86 0.86
1 17,598 0.86 0.85 0.86
XLNet 0 17,573 0.86 0.88 0.84 0.86
1 17,598 0.84 0.88 0.86
RoBERTa 0 17,573 0.87 0.87 0.86 0.86
1 17,598 0.86 0.87 0.87
XLM-RoBERTa 0 17,573 0.86 0.86 0.86 0.86
1 17,598 0.86 0.86 0.86
SBERT + SVM 0 17,573 0.85 0.84 0.87 0.86
1 17,598 0.87 0.83 0.85
SBERT +RF 0 17,573 0.81 0.79 0.86 0.82
1 17,598 0.84 0.77 0.81
TF-IDF + SVM 0 17,573 0.84 0.84 0.86 0.85
1 17,598 0.86 0.83 0.85
TF-IDF + RF 0 17,573 0.84 0.80 0.90 0.85
1 17,598 0.88 0.78 0.83

5. Conclusions
In this study, we aimed to ascertain a state-of-the-art (SOTA) language model for
automated cyberbullying (CB) detection. We prepared a new dataset (D2) from existing CB
studies. The datasets were originated from FormSpring and Twitter, and were manually
annotated. We used the dataset in our implementation, and our results showed RoBERTa
performed well in both experiments, cases 1 and 2. For a classification task, we argue that
large language models (LLMs) can predict the minority class better than the approach using
the traditional machine learning approach and/or the oversampling technique (case 1).
This is due to the ability of the language model to understand the context of long and short
text. In addition, we showed that RoBERTa perform better than deep learning approaches
like BiLSTM with an attention mechanism. We also evidenced that, when the dataset is
balanced, the traditional machine learning approach produces good performance; however,
RoBERTa yielded a state-of-the-art (SOTA) performance. To conclude, our contributions
can be summarised as follows. We prepared a new dataset (D2) for the development of
algorithms in the field of CB detection. The dataset (D2) has been made publicly available
for research access and use. We demonstrated how large language models can be used for
automated CB detection with two datasets (D1 and D2). We presented SOTA results for CB
detection by fine tuning RoBERTa.
In theory, the use of machine learning algorithms yields poor performance when
fed with imbalanced datasets compared to large language models. Similarly, language
models yield better results with balanced datasets. This implies that the performance
of RoBERTa is consistent across different categorises (balanced or not) of cyberbullying
datasets. In practice, our application is useful for social network owners, the government,
and developers to implement cyberbullying detection algorithms to prevent and reduce the
act. It is worth mentioning that our experiment is limited to English text. For future work,
our implementation can be extended to other languages; most especially, the LLMs can be
tuned with external non-English corpus for improving the SOTA models in non-English
contexts. In addition, we consider implementing a multimodal approach to develop and
enhance algorithms for CB detection. Furthermore, the implementation can be adopted to
detect other forms of online abuse, including hate speech and cyber molestation.
Analytics 2023, 2 705

Author Contributions: Conceptualization, B.O. and B.D.; methodology, B.O. and B.D.; software, B.D.;
validation, B.O. and B.D.; formal analysis, B.D.; investigation, B.O. and B.D.; resources B.D.; data
curation, B.O. and B.D.; writing—original draft preparation, B.O. and B.D.; writing—review and
editing, B.O.; visualization, B.O. and B.D.; supervision, B.O.; project administration, B.O.; All authors
have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: The dataset and code used in this work is available on the Github
repository (please see the link below) GitHub—Babitha23/Cyberbullying-detection Case 1: https:
//github.com/Babitha23/Cyberbullying-detection/tree/main/Case1. Case 2: https://github.com/
Babitha23/Cyberbullying-detection/tree/main/Case2.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Pew Research Center. A Majority of Teens Have Experienced Some Form of Cyberbullying. 2018. Available online: https:
//www.pewresearch.org/internet/2018/09/27/a-majority-of-teens-have-experienced-some-form-ofcyberbullying/ (accessed
on 16 March 2023).
2. Statista. Share of Adult Internet Users in the United States Who Have Personally Experienced Online Harassment as of January
2021. 2021. Available online: https://www.statista.com/statistics/333942/us-internet-online-harassment-severity/ (accessed on
16 March 2023).
3. Pew Research Center. Teens and Cyberbullying. 2022. Available online: https://www.pewresearch.org/internet/2022/12/15
/teens-and-cyberbullying-2022/ (accessed on 16 March 2023).
4. Office for National Statistics. Online Bullying in England and Wales; Year Ending March 2020. 2020. Available online:
https://www.ons.gov.uk/peoplepopulationandcommunity/crimeandjustice/bulletins/onlinebullyinginenglandandwales/
yearendingmarch2020 (accessed on 16 March 2023).
5. Patchin, J.W.; Hinduja, P.D.S. Tween Cyberbullying; Cyberbullying Research Center: Jupiter, FL, USA, 2020.
6. Emmery, C.; Verhoeven, B.; De Pauw, G.; Jacobs, G.; Van Hee, C.; Lefever, E.; Daelemans, W. Current limitations in cyberbullying
detection: On evaluation criteria, reproducibility, and data scarcity. Lang. Resour. Eval. 2021, 55, 597–633. [CrossRef]
7. Dinakar, K.; Reichart, R.; Lieberman, H. Modeling the detection of textual cyberbullying. In Proceedings of the International
AAAI Conference on Web and Social Media, Barcelona, Spain, 21 July 2011; Volume 5, pp. 11–17.
8. Reynolds, K.; Kontostathis, A.; Edwards, L. Using machine learning to detect cyberbullying. In Proceedings of the 10th
International Conference on Machine Learning and Applications and Workshops, Honolulu, HI, USA, 18–21 December 2011;
IEEE: Piscataway, NJ, USA, 2011; Volume 2, pp. 241–244.
9. Aboujaoude, E.; Savage, M.W.; Starcevic, V.; Salame, W.O. Cyberbullying: Review of an old problem gone viral. J. Adolesc. Health
2015, 57, 10–18. [CrossRef]
10. Xu, J.M.; Jun, K.S.; Zhu, X.; Bellmore, A. Learning from bullying traces in social media. In Proceedings of the 2012 Conference of
the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Montreal, QC,
Canada, 3–8 June 2012; pp. 656–666.
11. Huang, J.; Zhong, Z.; Zhang, H.; Li, L. Cyberbullying in social media and online games among Chinese college students and its
associated factors. Int. J. Environ. Res. Public Health 2021, 18, 4819. [CrossRef]
12. Hellfeldt, K.; López-Romero, L.; Andershed, H. Cyberbullying and psychological well-being in young adolescence: The potential
protective mediation effects of social support from family, friends, and teachers. Int. J. Environ. Res. Public Health 2020, 17, 45.
[CrossRef] [PubMed]
13. Nixon, C.L. Current perspectives: The impact of cyberbullying on adolescent health. Adolesc. Health Med. Ther. 2014, 5, 143–158.
[CrossRef] [PubMed]
14. Jin, X.; Zhang, K.; Twayigira, M.; Gao, X.; Xu, H.; Huang, C.; Luo, X.; Shen, Y. Cyberbullying among college students in a Chinese
population: Prevalence and associated clinical correlates. Front. Public Health 2023, 11, 1100069. [CrossRef]
15. Karki, A.; Thapa, B.; Pradhan, P.M.S.; Basel, P. Depression, anxiety, and stress among high school students: A cross-sectional
study in an urban municipality of Kathmandu, Nepal. PLoS Glob. Public Health 2022, 2, e0000516. [CrossRef]
16. Piccoli, V.; Carnaghi, A.; Bianchi, M.; Grassi, M. Perceived-Social Isolation and Cyberbullying Involvement: The Role of Online
Social Interaction. Int. J. Cyber Behav. Psychol. Learn. (IJCBPL) 2022, 12, 1–14. [CrossRef]
17. Peng, Z.; Klomek, A.B.; Li, L.; Su, X.; Sillanmäki, L.; Chudal, R.; Sourander, A. Associations between Chinese adolescents subjected
to traditional and cyber bullying and suicidal ideation, self-harm and suicide attempts. BMC Psychiatry 2019, 19, 324. [CrossRef]
18. Kim, S.; Kimber, M.; Boyle, M.H.; Georgiades, K. Sex differences in the association between cyberbullying victimization and
mental health, substance use, and suicidal ideation in adolescents. Can. J. Psychiatry 2019, 64, 126–135. [CrossRef]
Analytics 2023, 2 706

19. Zaborskis, A.; Ilionsky, G.; Tesler, R.; Heinz, A. The association between cyberbullying, school bullying, and suicidality among
adolescents. Crisis 2018, 40, 100–114. [CrossRef] [PubMed]
20. Islam, M.I.; Yunus, F.M.; Kabir, E.; Khanam, R. Evaluating risk and protective factors for suicidality and self-harm in Australian
adolescents with traditional bullying and cyberbullying victimizations. Am. J. Health Promot. 2022, 36, 73–83. [CrossRef] [PubMed]
21. Eyuboglu, M.; Eyuboglu, D.; Pala, S.C.; Oktar, D.; Demirtas, Z.; Arslantas, D.; Unsal, A. Traditional school bullying and
cyberbullying: Prevalence, the effect on mental health problems and self-harm behavior. Psychiatry Res. 2021, 297, 113730.
[CrossRef] [PubMed]
22. Messias, E.; Kindrick, K.; Castro, J. School bullying, cyberbullying, or both: Correlates of teen suicidality in the 2011 CDC youth
risk behavior survey. Compr. Psychiatry 2014, 55, 1063–1068. [CrossRef] [PubMed]
23. Elsafoury, F.; Katsigiannis, S.; Pervez, Z.; Ramzan, N. When the timeline meets the pipeline: A survey on automated cyberbullying
detection. IEEE Access 2021, 9, 103541–103563. [CrossRef]
24. Rosa, H.; Matos, D.; Ribeiro, R.; Coheur, L.; Carvalho, J.P. A “deeper” look at detecting cyberbullying in social networks. In
Proceedings of the 2018 International Joint Conference on Neural Networks (IJCNN), Rio de Janeiro, Brazil, 8–13 July 2018; IEEE:
Piscataway, NJ, USA, 2018; pp. 1–8.
25. Rosa, H.; Pereira, N.; Ribeiro, R.; Ferreira, P.; Carvalho, J.; Oliveira, S.; Coheur, L.; Paulino, P.; Simão, A.V.; Trancoso, I. Automatic
cyberbullying detection: A systematic review. Comput. Hum. Behav. 2019, 93, 333–345. [CrossRef]
26. Ogunleye, B.O. Statistical Learning Approaches to Sentiment Analysis in the Nigerian Banking Context. Ph.D. Thesis, Sheffield
Hallam University, Sheffield, UK, October 2021.
27. Yi, P.; Zubiaga, A. Cyberbullying detection across social media platforms via platform-aware adversarial encoding. In Proceedings
of the International AAAI Conference on Web and Social Media, Atlanta, GA, USA, 6–9 June 2022; Volume 16, pp. 1430–1434.
28. Agrawal, S.; Awekar, A. Deep learning for detecting cyberbullying across multiple social media platforms. In Advances in
Information Retrieval, Proceedings of the 40th European Conference on IR Research, ECIR 2018, Grenoble, France, 26–29 March 2018;
Springer International Publishing: Cham, Switzerland, 2018; pp. 141–153.
29. Di Capua, M.; Di Nardo, E.; Petrosino, A. Unsupervised cyber bullying detection in social networks. In Proceedings of the 23rd
International Conference on Pattern Recognition (ICPR), Cancun, Mexico, 4–8 December 2016; IEEE: Piscataway, NJ, USA, 2016;
pp. 432–437.
30. Kontostathis, A.; Reynolds, K.; Garron, A.; Edwards, L. Detecting cyberbullying: Query terms and techniques. In Proceedings of
the 5th Annual ACM Web Science Conference, Paris, France, 2–4 May 2013; pp. 195–204.
31. Centers for Disease Control and Prevention. Technology and Youth: Protecting Your Child from Electronic Aggression; Centers for
Disease Control and Prevention: Atlanta, GA, USA, 2014. Available online: http://www.cdc.gov/violenceprevention/pdf/ea-
tipsheet-a.pdf (accessed on 16 March 2023).
32. Waseem, Z.; Hovy, D. Hateful symbols or hateful people? Predictive features for hate speech detection on twitter. In Proceedings
of the NAACL Student Research Workshop, San Diego, CA, USA, 13–15 June 2016; pp. 88–93.
33. Wulczyn, E.; Thain, N.; Dixon, L. Ex machina: Personal attacks seen at scale. In Proceedings of the 26th International Conference
on World Wide Web, Perth, Australia, 3–7 April 2017; pp. 1391–1399.
34. Huang, H.; Qi, D. Cyberbullying detection on social media. High. Educ. Orient. Stud. 2023, 3, 74–86.
35. Alduailaj, A.M.; Belghith, A. Detecting Arabic Cyberbullying Tweets Using Machine Learning. Mach. Learn. Knowl. Extr. 2023, 5,
29–42. [CrossRef]
36. Lepe-Faúndez, M.; Segura-Navarrete, A.; Vidal-Castro, C.; Martínez-Araneda, C.; Rubio-Manzano, C. Detecting Aggressiveness
in Tweets: A Hybrid Model for Detecting Cyberbullying in the Spanish Language. Appl. Sci. 2021, 11, 10706. [CrossRef]
37. Dewani, A.; Memon, M.A.; Bhatti, S.; Sulaiman, A.; Hamdi, M.; Alshahrani, H.; Shaikh, A. Detection of Cyberbullying Patterns
in Low Resource Colloquial Roman Urdu Microtext Using Natural Language Processing, Machine Learning, and Ensemble
Techniques. Appl. Sci. 2023, 13, 2062. [CrossRef]
38. Suhas-Bharadwaj, R.; Kuzhalvaimozhi, S.; Vedavathi, N. A Novel Multimodal Hybrid Classifier Based Cyberbullying Detection
for Social Media Platform. In Data Science and Algorithms in Systems, Proceedings of 6th Computational Methods in Systems and
Software, Online, 12–15 October 2022; Springer International Publishing: Cham, Switzerland, 2022; Volume 2, pp. 689–699.
39. Woo, W.H.; Chua, H.N.; Gan, M.F. Cyberbullying Conceptualization, Characterization and Detection in social media—A
Systematic Literature Review. Int. J. Perceptive Cogn. Comput. 2023, 9, 101–121.
40. Ogunleye, B.; Maswera, T.; Hirsch, L.; Gaudoin, J.; Brunsdon, T. Comparison of Topic Modelling Approaches in the Banking
Context. Appl. Sci. 2023, 13, 797. [CrossRef]
41. Zhao, A.; Yu, Y. Knowledge-enabled BERT for aspect-based sentiment analysis. Knowl.-Based Syst. 2021, 227, 107220. [CrossRef]
42. Yang, N.; Jo, J.; Jeon, M.; Kim, W.; Kang, J. Semantic and explainable research-related recommendation system based on
semi-supervised methodology using BERT and LDA models. Expert Syst. Appl. 2022, 190, 116209. [CrossRef]
43. Lin, S.Y.; Kung, Y.C.; Leu, F.Y. Predictive intelligence in harmful news identification by BERT-based ensemble learning model
with text sentiment analysis. Inf. Process. Manag. 2022, 59, 102872. [CrossRef]
44. Paul, S.; Saha, S. CyberBERT: BERT for cyberbullying identification: BERT for cyberbullying identification. Multimed. Syst. 2022,
28, 1897–1904. [CrossRef]
Analytics 2023, 2 707

45. Yadav, J.; Kumar, D.; Chauhan, D. Cyberbullying detection using pre-trained bert model. In Proceedings of the 2020 International
Conference on Electronics and Sustainable Communication Systems (ICESC), Coimbatore, India, 2–4 July 2020; IEEE: Piscataway,
NJ, USA, 2020; pp. 1096–1100.
46. Devlin, J.; Chang, M.W.; Lee, K.; Toutanova, K. Bert: Pre-training of deep bidirectional transformers for language understanding.
arXiv 2018, arXiv:1810.04805.
47. Liu, Y.; Ott, M.; Goyal, N.; Du, J.; Joshi, M.; Chen, D.; Stoyanov, V. Roberta: A robustly optimized bert pretraining approach. arXiv
2019, arXiv:1907.11692.
48. Yang, Z.; Dai, Z.; Yang, Y.; Carbonell, J.; Salakhutdinov, R.R.; Le, Q.V. Xlnet: Generalized autoregressive pretraining for language
understanding. Adv. Neural Inf. Process. Syst. 2019, 32. [CrossRef]
49. Conneau, A.; Khandelwal, K.; Goyal, N.; Chaudhary, V.; Wenzek, G.; Guzmán, F.; Stoyanov, V. Unsupervised cross-lingual
representation learning at scale. arXiv 2019, arXiv:1911.02116.
50. Vapnik, V. The Nature of Statistical Learning Theory; Springer Science & Business Media: Berlin/Heidelberg, Germany, 1999.
51. Menon, R.V.; Chakrabarti, I. Low complexity VLSI architecture for improved primal–dual support vector machine learning core.
Microprocess. Microsyst. 2023, 98, 104806. [CrossRef]
52. Breiman, L. Random forests. Mach. Learn. 2001, 45, 5–32. [CrossRef]
53. Cao, M.; Yin, D.; Zhong, Y.; Lv, Y.; Lu, L. Detection of geochemical anomalies related to mineralization using the Random Forest
model optimized by the Competitive Mechanism and Beetle Antennae Search. J. Geochem. Explor. 2023, 249, 107195. [CrossRef]
54. Dinh, T.P.; Pham-Quoc, C.; Thinh, T.N.; Do Nguyen, B.K.; Kha, P.C. A flexible and efficient FPGA-based random forest architecture
for IoT applications. Internet Things 2023, 22, 100813. [CrossRef]
55. Koohmishi, M.; Azarhoosh, A.; Naderpour, H. Assessing the key factors affecting the substructure of ballast-less railway track
under moving load using a double-beam model and random forest method. In Structures; Elsevier: Amsterdam, The Netherlands,
2023; Volume 55, pp. 1388–1405.
56. Wang, J.; Fu, K.; Lu, C.T. Sosnet: A graph convolutional network approach to fine-grained cyberbullying detection. In Proceedings
of the 2020 IEEE International Conference on Big Data (Big Data), Atlanta, GA, USA, 10–13 December 2020; IEEE: Piscataway, NJ,
USA, 2020; pp. 1699–1708.
57. Bretschneider, U.; Wöhner, T.; Peters, R. Detecting online harassment in social networks. In Proceedings of the Thirty Fifth
International Conference on Information Systems, Auckland, New Zealand, 14–17 December 2014.
58. Chatzakou, D.; Leontiadis, I.; Blackburn, J.; Cristofaro, E.D.; Stringhini, G.; Vakali, A.; Kourtellis, N. Detecting cyberbullying and
cyberaggression in social media. ACM Trans. Web (TWEB) 2019, 13, 17. [CrossRef]
59. Davidson, T.; Warmsley, D.; Macy, M.; Weber, I. Automated hate speech detection and the problem of offensive language. In
Proceedings of the International AAAI Conference on Web and Social Media, Montreal, QC, Canada, 11 March 2017; Volume 11,
pp. 512–515.
60. Reimers, N.; Gurevych, I. Sentence-bert: Sentence embeddings using siamese bert-networks. arXiv 2019, arXiv:1908.10084.

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.