=Paper=
{{Paper
|id=Vol-3681/T6-14
|storemode=property
|title=Examining Hate Speech Detection Across Multiple Indo-Aryan Languages in Tasks 1 & 4
|pdfUrl=https://ceur-ws.org/Vol-3681/T6-14.pdf
|volume=Vol-3681
|authors=Gyandeep Kalita,Eisha Halder,Chetna Taparia,Advaitha Vetagiri,Dr. Partha Pakray
|dblpUrl=https://dblp.org/rec/conf/fire/KalitaHTVP23
}}
==Examining Hate Speech Detection Across Multiple Indo-Aryan Languages in Tasks 1 & 4==
https://ceur-ws.org/Vol-3681/T6-14.pdf
Examining Hate Speech Detection Across Multiple
Indo-Aryan Languages in Tasks 1 & 4
Gyandeep Kalita1,† , Eisha Halder1,† , Chetna Taparia1,† , Advaitha Vetagiri1,* and
Dr. Partha Pakray1
1
National Institute of Technology Silchar
Abstract
Hate speech continues to be a pressing concern in online social media (OSM) platforms, necessitating
effective automated detection systems. In this paper, we propose a unified approach, encompassing both
Task 1 & 4, to tackle the challenge of hate speech recognition within the HASOC 2023 framework. It
addresses the complexities of multilingual OSM by employing cutting-edge Natural Language Processing
(NLP) techniques and leveraging powerful language models put forward by team CNLP-NITS-PP. The
key objective is optimising precision-recall trade-offs in hate speech detection, spanning English and
Indo-Aryan languages. The empirical results demonstrate the effectiveness of our approach in isolating
explicit signs of hate speech, emphasizing model efficiency, interpretability, and the importance of
diverse linguistic nuances in creating safer online environments. This integrated work sets the stage
for advancements in hate-span detection and underlines the significance of fostering responsible and
inclusive online conversations across various language environments.
Keywords
Online social media, Multilingual, Natural Language Processing, CNN, BiLSTM, BERT, GPT-2, Named
Entity Recognition.
1. Introduction
Social media platforms such as Twitter and Facebook have become integral to modern life,
providing a global platform for individuals to express themselves. However, the openness of
these platforms has also led to the proliferation of harmful content, including hate speech and
harassment [1]. This has underscored the need for automated systems to identify and address
abusive language in online conversations [2] [3].
Detecting offensive content is challenging due to its diverse linguistic forms, necessitating
context-aware models to pinpoint hateful or abusive text snippets [4]. Additionally, implicit
forms of hate speech require the deduction of pragmatic implications [5].
The spread of hate speech and inflammatory language on social media platforms is a major
worldwide problem in today’s digital age, as communication plays a crucial role in determining
Forum for Information Retrieval Evaluation, December 15-18, 2023, India
*
Corresponding author.
†
These authors contributed equally for Task 1 & 4.
$ gyandeepkalita1@gmail.com (G. Kalita); eishashalder@gmail.com (E. Halder); chetna.taparia@gmail.com
(C. Taparia); advaitha21\protect1_rs@cse.nits.ac.in (A. Vetagiri); partha@cse.nits.ac.in (Dr. P. Pakray)
https://parthapakray.com/ (Dr. P. Pakray)
0000-0002-0651-4171 (A. Vetagiri); 0000-0003-3834-5154 (Dr. P. Pakray)
© 2023 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR
Workshop
Proceedings
http://ceur-ws.org
ISSN 1613-0073
CEUR Workshop Proceedings (CEUR-WS.org)
CEUR
ceur-ws.org
Workshop ISSN 1613-0073
Proceedings
public debate. Low-resource languages like Sinhala, Gujarati, Bengali, Bodo, and Assamese,
which have received little attention in the field of Natural Language Processing (NLP), are
severely affected by this problem.
Our research activities cover a range of tasks for identifying harmful and hateful content in
these underrepresented languages1 . In Task 1, we tackle Sinhala (Task 1A), a language with
a unique alphabet and intricate grammatical structures, and further broaden our emphasis to
Gujarati (Task 1B), where a dearth of labelled data poses a significant obstacle. Task 4 extends
the study’s horizons by including Bengali, Bodo, and Assamese [6] [7]. These languages, which
are rich in cultural richness and legacy, have generally been disregarded in NLP research,
especially when it comes to the identification of hate speech. Our study uses statistics painstak-
ingly gathered from social media sites to use binary classification to characterize material as
hate/offensive or not.
The importance of our work lies in its role in safeguarding cultural identities and developing
secure online environments for these language speakers. To respond to the complexities of
hate speech in different linguistic and cultural contexts, we use cutting-edge NLP approaches,
language-specific feature engineering, and pre-processing. Additionally, we investigate how
models developed for languages with abundant resources may be applied to languages with
limited resources to improve hate speech identification.
Through this extensive study project, we hope to advance responsible digital communication,
better understand how to identify hate speech in different linguistic contexts, and create a more
welcoming online space for every language.
2. Literature Review
Hate speech detection in fairly low-resourced languages such as Sinhala and regional Indian
languages has recently attracted research attention [8]. With the proliferation of user-generated
content on social media platforms, there is an urgent need to identify and moderate hateful and
offensive content in these regional languages (Mathew et al., 2021)[9].
For the Sinhala Language, a few research works have been conducted. (Munasinghe et al.,
2022)[10] contributed an annotated dataset of Sinhala Tweets annotated into Hate or Non-Hate.
They also developed and compared the performance of different architectures such as CNN,
LSTM, BiGRU and an ensemble of various other Deep Learning architectures. (Sandaruwan et
al., 2019)[11] also contributed a labeled dataset containing texts from Facebook and YouTube for
Hate Detection in Sinhala and compared classification results using simple Machine Learning
Classifiers such as SVMs, MNB, RFDT, etc. The SOLD: Sinhala Offensive Language Dataset, a
labeled dataset for Offensive content detection in Sinhala, was contributed by (Ranasinghe et
al.,2022)[12], which also forms the basis for the dataset provided for the Task 1A of HASOC
2023.
In case of Indian Languages, prior work on hate speech detection has concentrated primarily
on Hindi and Malayalam. For instance, (Bohra et al., 2018)[13] presented a dataset for hate speech
identification in Hindi-English code-mixed social media text. They tested various classification
models, including fastText, CNN, GRU, and LSTM
1
Github Repository
For Gujarati, (Khurana et al., 2022)[14] contributed a novel model to detect hate comments in
13 Indian languages that included Gujarati based on XLM-RoBERTa (XLM-R) using the Moj
Multilingual Abusive Comment Identification dataset.
For the Bengali language, (Das et al., 2020)[15] compiled a labelled dataset of YouTube
comments for Bengali hate speech recognition. They compared machine learning models like
SVM, NB and deep learning architectures like CNN, GRU, and capsule networks.
Assamese is a relatively low-resource language. (Ghosh et al., 2023)[16] contributed a dataset
for binary hate classification in Assamese and described an approach for hate detection using
various BERT models.
For the Bodo language, there is limited prior research. The HASOC 2023 shared task provides
the pioneering benchmark hate speech detection dataset in Bodo. This will encourage further
research in this low-resource language (Chakravarthi et al., 2021)[17]
(Vetagiri et al., 2023a)[18] leveraged GPT-2 to automatically classify online sexist content.
Their work demonstrates the potential of sizable pre-trained language models for hate speech
detection. In another work, (Vetagiri et al., 2023b)[4] proposed an approach using CNN-BiLSTM
and domain-specific embeddings for online sexism prediction.
Much previous work has relied on machine learning and deep neural networks. But these
necessitate substantial labelled datasets, which are scarce for low-resource languages. Recent
emphasis has focused on multilingual models such as mBERT, which can leverage data from
high-resource languages. Domain adaptation approaches have also proven effective in adapting
models trained on English data.
The HASOC 2023 shared tasks furnishes standard labeled benchmark datasets for hate speech
detection. This will catalyze research in these languages and progress the state-of-the-art.
Multilingual models and cross-lingual transfer learning are promising avenues to explore for
these languages.
3. Dataset and Task Description
3.1. Tasks Description:
Task 1 of the HASOC’23 aimed at identifying hate, offensive, and profane content in social
media posts in two languages, namely Sinhala(Task 1a) and Gujarati(Task 1b) [19].
Task 4 was similar to Task 1 and required us to detect hate speech in three other Indian
languages, Bengali, Assamese and Bodo. For all the given languages, the training and test
datasets had already been provided.
Creating coarse-grained binary classification models to divide tweets into the following two
categories was the primary goal for the tasks:
• Hate and Offensive(HOF): Posts that contain hate speech, vulgarity, or offensive material.
• Non-Hate and Offensive (NOT): Posts devoid of offensive language, hate speech, or any
other negative elements.
3.2. Data Source
3.2.1. Sinhala Dataset (Task 1a)
The Sinhala dataset provided for train and test had been sourced from the recently released
SOLD: Sinhala Offensive Language Detection dataset, which served as a comprehensive resource
for the particular task. The training dataset had been further divided into three columns. The
first one consisted of the post id, the second of the tweet text, while the third column consisted
of the labels, HOF and NOT, for each of the corresponding tweets in the same row.
3.2.2. Gujarati Dataset(Task 1b)
For Task 1b, the training dataset had 200 tweets, primarily categorized into two labels, HOF
and NOT, besides three other columns, including the tweet id, the UserName and the date of
creation. It is noteworthy that the exact source of the dataset has not been mentioned in the
materials provided for the competition.
3.2.3. Assamese, Bengali and Bodo Dataset(Task 4)
The training and test datasets for the task had already been provided in all three languages,
Assamese Bengali and Bodo. However, it is worth noting that none of the sources for the data
were explicitly mentioned.
3.3. Data Statistics
Distribution of HASOC’23 training datasets for Task 1 and Task 4. For each language, the total
no of text entries and the corresponding no of tweets per class are shown below.
Table 1
Task 1 & 4 dataset stastistics.
Language Total text entries Hate and Offensive(HOT) Not Hate(NOT)
Assamese 4035 2346 1689
Bengali 2180 515 1665
Bodo 1678 998 680
Sinhala 7500 3176 4324
Gujarati 200 100 100
Besides this, any external use of data beyond what was provided required explicit permission
from the competition organizers.
Test Set size:
For Task 1A, the Sinhala test dataset consisted of 2500 tweets. This had to be labeled as either
HOF or NOT based on our model.
For Task 1B, the Gujarati test dataset consisted of a total of 1196 tweets to be labelled similarly.
For Task 4, the Bengali, Assamese, and Bodo test datasets consisted of 320, 1009, and 420 text
entries, respectively. These entries had to be labeled as either HOF or NOT based on our model.
3.4. Data Preprocessing
We used a number of standard preprocessing methods prior to training our model using the
given datasets. Given that the training datasets provided had their texts sourced from Twitter,
it was anticipated to contain certain unwanted elements, such as emojis, URLs, mentions and
special characters. In order to guarantee the accuracy and relevancy of the text data, we followed
procedures to remove such unwanted noise.
4. Methodology
In this section, we describe the methodology and the experimental setup used for the various
tasks under HASOC 2023. We conducted a thorough investigation into various neural network
architectures, pretrained Large Language models, and classical machine learning models to
identify the most effective model for the task.
4.1. Task 1: Identifying Hate, offensive and profane content in Sinhala &
Gujarati
For the task of Identifying Hate, offensive and profane content in Sinhala & Gujarati, the models
that resulted in the best performance are as follows:
• A CNN-based Binary Classification Model with FastText Embeddings.
• A CNN-BiLSTM based Hybrid Model with FastText/GloVE Embeddings
4.1.1. CNN + FastText Binary Classification Model :
Inspired by the works of Kim et al., [20], we developed the model based on the CNN architecture.
At the core of our model lies the input layer, where text sequences representing individual posts
are processed. To prepare the input data, we concatenate the words within each sentence, with
the sequence length capped at 70 words. The words here are represented as dense vectors of
300 dimensions using pre-trained FastText embeddings for the respective languages. Using
machine learning or related dimensional reduction techniques, word embedding converts each
token into a vector of real numbers to quantify and classify the semantic similarity of linguistic
phrases based on their distributional qualities in a large corpus.
For the convolutional layer, we employed a one-dimensional convolution operation utilizing
100 filters with a kernel size of 3, leading to a systematic scanning of the text sequences
and identifying pertinent patterns in the data. An activation function, the rectified linear
unit (ReLU), was also applied to introduce non-linearity and enable the model to capture
complex relationships in the data. Subsequently, a dense layer with 50 neurons and a ReLU
activation function, coupled with an L2-norm constraint, was added to transform the extracted
features further. Dropout with a rate of 0.5 was applied as a regularization technique to prevent
overfitting. The resultant vector was then concatenated with the feature vector and the output
was passed onto a dense output layer with sigmoid activation and cross entropy loss as shown
in Figure 1, to produce the binary hate classification for the model.
4.1.2. CNN-BiLSTM + FastText/GLoVE Binary Classification Model :
Based on the contributions of Vetagiri et al., [4], we developed the model, which is a combination
of two different model architectures - the Convolutional Neural Networks (CNN) (Kim, 2014)[20]
layer for identifying local textual patterns in the input text and Bidirectional Long Short-Term
Memory (BiLSTM) Liu and Guo et al., [21] layer as a form of the Recurrent Neural Architecture
Sherstinsky et al., [22] for understanding the long-term complex sequential dependencies within
the text data.
The output of these two layers is then passed through a dense layer with a sigmoid activation
function for Binary Classification. To prepare the Input data, the model uses the exact pre-
trained FastText embeddings for the respective language mentioned above, representing the
words as 300-dimensional dense vectors, with the sequence length capped at 70 words, which
was held as non-trainable. A similar implementation of this model using pre-trained GLoVE
embeddings Kumar et al., [23] showed identical results. For the CNN layers, we first employed a
SpatialDropout1D layer, a dropout variant that selectively drops entire 1D feature maps during
training, to combat overfitting. Subsequently, a one-dimensional convolution layer with 64
filters and a kernel size of 3 was used to capture local textual patterns with fine granularity.
For the BiLSTM part, we used the initial layer with 128 units and a return sequence setting
with a dropout of 0.1 and recurrent dropout of 0.1 to process the text inputs in both forward as
well as reverse directions followed by a Global Average Pooling and a dense layer with 128 units
and a rectified linear unit (ReLU) activation function to introduce non-linearity. Subsequently,
a dropout layer is employed whose output is then concatenated with the feature vector and
passed through a dense layer with sigmoid activation as shown in figure 2 to produce the overall
model for Binary Hate Classification. Our models are trained using the RMSprop optimiser, and
our loss function is a binary cross-entropy function. To fine-tune our hyper-parameters over a
range of values, we conduct a grid search and select the best-performing model according to
validity accuracy. However, no attempt at experimentation by reversing the order of the CNN
& Bi-LSTM Layers was made for this particular task.
4.2. Task 4: Identifying Hate, offensive and profane content in Bengali, Bodo,
and Assamese languages
The two model architectures used in Task 1 were also experimented with in Task 4. These
architectures were used to implement the CNN and the CNN-BiLSTM models which used
pre-trained FastText embeddings in the respective languages (except for Bodo, for which Hindi
embeddings were used) representing each word as a dense vector with 300 dimensions.
In addition to these two architectures, several others were also experimented with, the details
for which are discussed below:
Figure 1: CNN + FastText Binary Classification Figure 2: CNN-BiLSTM + FastText/GLoVE Binary
Model Architecture Classification Model Architecture
4.2.1. Pre-trained BERT Architecture
Considering the low-resource nature of Task 4 and the limited size of the datasets, we exper-
imented with pre-trained models based on the Bidirectional Encoder Representations from
Transformers (BERT) architecture Devlin et al., [24] [25]. We experimented with the Tensorflow
Hub to access the pre-trained BERT models.
We used the "bert-multi-cased-preprocess/3" for text processing and the "bert-multi-cased-L-
12-H-768-A-12/4" encoder for contextualized word embeddings from the TensorFlow Hub, which
is trained on multilingual Wikipedia Data. The model utilizes a BERT preprocessing layer for
tokenization and embeddings of the input text, followed by a BERT encoder layer to generate
contextual embeddings containing the complex contextual relationship within the language.
This is followed by a dropout Neural Layer with a 10% dropout rate to enhance generalization
and mitigate overfitting. The final trainable dense layer with 769 employs the sigmoid activation
function, producing the binary classification outputs for the given languages
4.2.2. GPT-2 Model
We also explored GPT-2 as a state-of-the-art pre-trained large language model for the task.
GPT-2 is a transformer-based model that takes a sequence of words, represented as dense
vectors, as input and uses many intermediate layers to extract contextual information for the
input text. The output is then passed through a dense layer, producing the Binary Classifier.
For the task, we used a pre-trained GPT-2 that contained 768 parameters, fine-tuned on the
training dataset for each language in an 80-20% split and a further 20% split from the training
set for validation. The input text is tokenised and passed through the model for fine-tuning.
Figure 3: Pre-trained BERT Figure 4: GPT-2 Model Architecture
Model Architecture
The model uses an Adam optimizer for optimization with a learning rate of 1e-5. The batch size
for the model is 8
4.2.3. Classical Machine-Learning Based Models
Due to the small size of the training datasets provided for all three languages, we tried implement-
ing classical Machine-Learning approaches as well to accomplish our goals. We experimented
with multiple Machine-Learning architectures such as Support Vector Machines (SVMs), Linear
Regression, Logistic Regression and Random Forest Classifiers.
However, although the training set accuracies and the macro F1 scores for all the above
architectures were quite close, we observed that the Logistic Regression Model achieved the best
overall performance. Hence, we created a simple Logistic Regression Model using SciKitLearn
Library, trained the Model with the training datasets for each language, and implemented simple
tf-idf vectorisation for embedding the input sentences
5. Results and Analysis
5.1. Task 1 & 4
For evaluation the models, the test accuracy for this test data was used for the initial evaluation
of the models. In addition to this, the Macro F1 scores, which were acquired by the models on
submission in the HASOC 2023 Submission portal were also considered. For Task 4, although
the performance of each of the aforementioned models was analyzed, only a few of the models
for each language gave the best performance, summarized in Tables 1 & 2.
As it is evident from Tables 1 & 2, the CNN-BiLSTM+FastText/GLoVe Model gave the best
Macro F1 Score for Task 1A(Sinhala), and the CNN+FastText Model gave the best Macro F1
Table 2
Performance scores for Tasks 1A & 1B in terms of their test accuracies and Macro F1 scores
Task Language Model Accuracy Macro F1
1A Sinhala CNN+FastText 0.7800 0.7556
CNN-BiLSTM+FastText 0.7781 0.7711
1B Gujarati CNN+FastText 0.7025 0.6873
CNN-BiLSTM+FastText 0.7121 0.6758
Table 3
Performance scores for Task 4 in terms of their test accuracies
Language Model Accuracy Macro F1
CNN+FastText 0.6381 0.60108
CNN-BiLSTM+FastText/GLoVe 0.6122
Bengali CNN+FastText+ External Dataset 0.8177
GPT-2 0.5924
BERT 0.5992
Logistic Regression 0.6325
CNN+FastText 0.6374 0.59485
Assamese CNN-BiLSTM+FastText/GLoVe 0.6183
GPT-2 0.5825
Logistic Regression 0.6559
CNN+FastText 0.6577
Bodo CNN-BiLSTM+FastText/GLoVe 0.6162
Logistic Regression 0.6755 0.66925
Score for Task 1B(Gujarati). For Task 4, in both Bengali and Assamese, the CNN+FastText Model
gave the highest accuracy with a Macro F1 score of 0.60108 in Bengali and 0.59485 in Assamese.
For Bodo, evidently, the simple Logistic Regression Model gave a much better performance than
the other model architectures with a Macro F1 score of 0.66925, possibly due to the small size of
the training dataset.
6. Conclusion and Future Scope
Our research has been dedicated to the vital task of identifying hate speech, particularly
in Indo-Aryan languages such as Bengali, Assamese, Bodo, Gujarati, and Sinhala. We’ve
devised a comprehensive strategy that unites these linguistic intricacies under a single versatile
model. Through domain-aware pre-training and meticulous alignment of our models with
language-specific context, we’ve significantly enhanced hate speech detection. Furthermore,
our exploration into model ensemble techniques has bolstered detection accuracy and resilience
across diverse language settings, laying a foundational step towards comprehensive hate speech
detection in Indo-Aryan languages. Our overarching goal is to foster a safer and more inclusive
digital space for speakers of diverse linguistic backgrounds.
As we look to the future, our work paves the way for further advancements in hate-span
detection, focusing on model efficiency, interpretability, and an expansive training data corpus
encompassing evolving hate speech trends and linguistic variations. We also recognize the
potential of real-time monitoring and context-aware integration in dynamically evolving online
environments.
Acknowledgments
We wish to extend our appreciation to the Computer Science and Engineering Department of
the National Institute of Technology Silchar for granting us the opportunity to carry out our
research and experiments. We are grateful for the support, resources, and research environment
offered by the CNLP & AI Lab at NIT Silchar.
References
[1] B. Vidgen, L. Derczynski, (2020), Directions in abusive language training data, a systematic
review: Garbage in, garbage out. PloS one 15 (2020).
[2] P. Fortuna, S. Nunes, A survey on automatic detection of hate speech in text, ACM
Computing Surveys (CSUR) 51 (2018) 1–30.
[3] S. Satapara, S. Masud, H. Madhu, M. A. Khan, M. S. Akhtar, T. Chakraborty, S. Modha,
T. Mandl, Overview of the HASOC subtracks at FIRE 2023: Detection of hate spans and
conversational hate-speech, in: Proceedings of the 15th Annual Meeting of the Forum
for Information Retrieval Evaluation, FIRE 2023, Goa, India. December 15-18, 2023, ACM,
2023.
[4] A. Vetagiri, P. Adhikary, P. Pakray, A. Das, CNLP-NITS at SemEval-2023 task 10: Online
sexism prediction, PREDHATE!, in: Proceedings of the 17th International Workshop on
Semantic Evaluation (SemEval-2023), Association for Computational Linguistics, Toronto,
Canada, 2023, pp. 815–822. URL: https://aclanthology.org/2023.semeval-1.113. doi:10.
18653/v1/2023.semeval-1.113.
[5] D. Jurgens, L. Hemphill, E. Chandrasekharan, A just and comprehensive strategy for using
NLP to address online abuse, Proceedings of the 57th Annual Meeting of the Association
for Computational Linguistics, 2019.
[6] K. Ghosh, A. Senapati, A. S. Pal, Annihilate Hates (Task 4, HASOC 2023): Hate Speech
Detection in Assamese, Bengali, and Bodo languages, in: Working Notes of FIRE 2023 -
Forum for Information Retrieval Evaluation, CEUR, 2023.
[7] T. Ranasinghe, K. Ghosh, A. S. Pal, A. Senapati, A. E. Dmonte, M. Zampieri, S. Modha,
S. Satapara, Overview of the HASOC subtracks at FIRE 2023: Hate speech and offensive
content identification in assamese, bengali, bodo, gujarati and sinhala, in: Proceedings of
the 15th Annual Meeting of the Forum for Information Retrieval Evaluation, FIRE 2023,
Goa, India. December 15-18, 2023, ACM, 2023.
[8] K. Ghosh, D. A. Senapati, Hate speech detection: a comparison of mono and multilingual
transformer model with cross-language evaluation, in: Proceedings of the 36th Pacific Asia
Conference on Language, Information and Computation, De La Salle University, Manila,
Philippines, 2022, pp. 853–865. URL: https://aclanthology.org/2022.paclic-1.94.
[9] B. Mathew, P. Saha, S. M. Yimam, C. Biemann, P. Goyal, A. Mukherjee, Hatexplain: A
benchmark dataset for explainable hate speech detection, in: Proceedings of the AAAI
conference on artificial intelligence, volume 35, 2021, pp. 14867–14875.
[10] S. Munasinghe, U. Thayasivam, A deep learning ensemble hate speech detection approach
for sinhala tweets, in: 2022 Moratuwa Engineering Research Conference (MERCon), 2022,
pp. 1–6. doi:10.1109/MERCon55799.2022.9906232.
[11] H. Sandaruwan, S. Lorensuhewa, M. Kalyani, Sinhala hate speech detection in social
media using text mining and machine learning, in: 2019 19th International Conference on
Advances in ICT for Emerging Regions (ICTer), volume 250, 2019, pp. 1–8. doi:10.1109/
ICTer48817.2019.9023655.
[12] T. Ranasinghe, I. Anuradha, D. Premasiri, K. Silva, H. Hettiarachchi, L. Uyangodage,
M. Zampieri, Sold: Sinhala offensive language dataset, 2022. arXiv:2212.00851.
[13] A. Bohra, D. Vijay, V. Singh, S. S. Akhtar, M. Shrivastava, A dataset of hindi-english
code-mixed social media text for hate speech detection, in: Proceedings of the second
workshop on computational modeling of people’s opinions, personality, and emotions in
social media, 2018, pp. 36–41.
[14] R. Khurana, C. Pandey, P. Gupta, P. Nagrath, AniMOJity:detecting hate comments in
Indic languages and analysing bias against content creators, in: Proceedings of the
19th International Conference on Natural Language Processing (ICON), Association for
Computational Linguistics, New Delhi, India, 2022, pp. 172–182. URL: https://aclanthology.
org/2022.icon-main.23.
[15] A. Das, N. Tandon, S. Narayan, HASoC: Hate Speech and Offensive Content Identification
in Indo-European Languages: Overview of HASoC Track at FIRE 2020, in: Forum for
Information Retrieval Evaluation (FIRE), 2020, pp. 29–32.
[16] K. Ghosh, D. Sonowal, A. Basumatary, B. Gogoi, A. Senapati, Transformer-based hate
speech detection in assamese, in: 2023 IEEE Guwahati Subsection Conference (GCON),
2023, pp. 1–5. doi:10.1109/GCON58516.2023.10183497.
[17] B. R. Chakravarthi, R. Priyadharshini, N. Jose, A. Kumar M, T. Mandl, P. K. Kumaresan,
R. Ponnusamy, H. R L, J. P. McCrae, E. Sherly, Findings of the shared task on offensive
language identification in Tamil, Malayalam, and Kannada, in: Proceedings of the First
Workshop on Speech and Language Technologies for Dravidian Languages, Association
for Computational Linguistics, Kyiv, 2021, pp. 133–145. URL: https://aclanthology.org/2021.
dravidianlangtech-1.17.
[18] A. Vetagiri, P. K. Adhikary, P. Pakray, A. Das, “Leveraging GPT-2 for Automated Classifi-
cation of Online Sexist Content“, In Exist 2023 Lab at CLEF 2023: Conference and Labs of
the Evaluation Forum, September 18–21, 2023, Thessaloniki, Greece, 2023.
[19] S. Satapara, H. Madhu, T. Ranasinghe, A. E. Dmonte, M. Zampieri, P. Pandya, N. Shah,
M. Sandip, P. Majumder, T. Mandl, Overview of the hasoc subtrack at fire 2023: Hate-
speech identification in sinhala and gujarati, in: K. Ghosh, T. Mandl, P. Majumder, M. Mitra
(Eds.), Working Notes of FIRE 2023 - Forum for Information Retrieval Evaluation, Goa,
India. December 15-18, 2023, CEUR Workshop Proceedings, CEUR-WS.org, 2023.
[20] Y. Kim, Convolutional neural networks for sentence classification, 2014.
arXiv:1408.5882.
[21] G. Liu, J. Guo, Bidirectional lstm with attention mechanism and convolutional
layer for text classification, Neurocomputing 337 (2019) 325–338. URL: https://
www.sciencedirect.com/science/article/pii/S0925231219301067. doi:https://doi.org/
10.1016/j.neucom.2019.01.078.
[22] A. Sherstinsky, Fundamentals of recurrent neural network (rnn) and long short-term
memory (lstm) network, Physica D: Nonlinear Phenomena 404 (2020) 132306. URL:
https://www.sciencedirect.com/science/article/pii/S0167278919305974. doi:https://doi.
org/10.1016/j.physd.2019.132306.
[23] S. Kumar, S. Kumar, D. Kanojia, P. Bhattacharyya, “a passage to India”: Pre-trained word
embeddings for Indian languages, in: Proceedings of the 1st Joint Workshop on Spoken
Language Technologies for Under-resourced languages (SLTU) and Collaboration and
Computing for Under-Resourced Languages (CCURL), European Language Resources
association, Marseille, France, 2020, pp. 352–357. URL: https://aclanthology.org/2020.sltu-1.
49.
[24] J. Devlin, M. Chang, K. Lee, K. Toutanova, BERT: pre-training of deep bidirectional
transformers for language understanding, CoRR abs/1810.04805 (2018). URL: http://arxiv.
org/abs/1810.04805. arXiv:1810.04805.
[25] K. Ghosh, A. Senapati, U. Garain, Baseline bert models for conversational hate speech
detection in code-mixed tweets utilizing data augmentation and offensive language identifi-
cation in marathi, in: Fire, 2022. URL: https://api.semanticscholar.org/CorpusID:259123570.