PreprintPDF Available
Preprints and early-stage research may not have been peer reviewed yet.

Abstract and Figures

The extensive rise in consumption of online social media (OSMs) by a large number of people poses a critical problem of curbing the spread of hateful content on these platforms. With the growing usage of OSMs in multiple languages, the task of detecting and characterizing hate becomes more complex. The subtle variations of code-mixed texts along with switching scripts only add to the complexity. This paper presents a solution for the HASOC 2021 Multilingual Twitter Hate-Speech Detection challenge by team PreCog IIIT Hyderabad. We adopt a multilingual transformer based approach and describe our architecture for all 6 sub-tasks as part of the challenge. Out of the 6 teams that participated in all the sub tasks, our submissions rank 3rd overall.
Content may be subject to copyright.
Battling Hateful Content in Indic Languages
HASOC ’21
Aditya Kadama,Anmol Goela,Jivitesh Jaina,Jushaan Singh Kalrab,
Mallika Subramaniana,Manvith Reddya,Prashant Kodalia,T.H. Arjuna,
Manish Shrivastavaaand Ponnurangam Kumaragurua
aInternational Institute of Information Technology, Hyderabad, India
bDelhi Technological University, Delhi, India
Abstract
The extensive rise in consumption of online social media (OSMs) by a large number of people poses a
critical problem of curbing the spread of hateful content on these platforms. With the growing usage
of OSMs in multiple languages, the task of detecting and characterizing hate becomes more complex.
The subtle variations of code-mixed texts along with switching scripts only add to the complexity. This
paper presents a solution for the HASOC 2021 Multilingual Twitter Hate-Speech Detection challenge
by team PreCog IIIT Hyderabad. We adopt a multilingual transformer based approach and describe our
architecture for all 6 sub-tasks as part of the challenge. Out of the 6 teams that participated in all the sub
tasks, our submissions rank 3rd overall.
Keywords
Hate Speech, Social Media, Code Mixed, Indic Languages, Transformer Architecture
1. Introduction
Dissemination of hateful content on nearly all social media is increasingly becoming an alarming
concern. In the research community as well, this is a heavily studied research problem [
1
,
2
,
3
,
4
,
5
]. Misconduct such as bullying, derogatory comments based on gender, race, religion,
threatening remarks etc. are more prevalent today than ever before. The repercussions that
such content can have is profound and can result in increased mental stress, emotional outburst
and negative psychological impacts [
6
]. Hence, curbing the proliferation of this hate speech is
imperative. Furthermore, the massive scale at which online social media platforms function
makes it an even more pressing issue, which needs to be addressed in a robust manner. Most
online social media platforms have imposed strict guidelines
123
to help prevent the spread
HASOC (2021) Hate Speech and Oensive Content Identication in English and Indo-Aryan Languages
"aditya.kadam@research.iiit.ac.in (A. Kadam); agoel00@gmail.com (A. Goel); jivitesh.jain@students.iiit.ac.in
(J. Jain); jushaan18@gmail.com (J. S. Kalra); mallika.subramanian@students.iiit.ac.in (M. Subramanian);
manvith.reddy@students.iiit.ac.in (M. Reddy); prashant.kodali@research.iiit.ac.in (P. Kodali);
arjun.thekoot@research.iiit.ac.in (T.H. Arjun); m.shrivastava@iiit.ac.in (M. Shrivastava); pk.guru@iiit.ac.in
(P. Kumaraguru)
©2020 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)
1https://help.twitter.com/en/rules-and-policies/hateful-conduct-policy
2https://transparency.fb.com/en-gb/policies/community-standards/hate-speech/
3https://support.google.com/youtube/answer/2801939
arXiv:2110.12780v1 [cs.CL] 25 Oct 2021
of hate. Inspite of these platform regulations, the dynamics of user-interaction inuence the
diusion of (and hence increase in) hate to a large extent [1].
The problem of hate speech has been addressed by several researchers, but the rise in multilin-
gual content has added to the complexity of identication of hateful content. Majority of these
studies deal with high-resource languages such as English, and only recently have low-resource
languages – such as several Indic Languages – been more deeply explored [
7
]. In a country
like India, with multitude of regional languages, the phenomenon of Code Mixing/Switching
(wherein linguistic units such as phrases/words of two languages occur in a single utterance) is
also pervasive.
In this paper we elucidate our approach in solving the six downstream tasks of hate speech
identication and characterization in Indian languages as a part of the ‘HASOC ’21 Hate Speech
and Oensive Content Identication in English and Indo-Aryan Languages’ challenge [
8
]. Moti-
vated by existing architectures, we curate our own pipeline by fusing ne-tuned transformer
based models with additional features to solve this challenge and highlight the dierent method-
ologies that were adopted for the three languages – English, Hindi, Marathi, and Code Mixed
Hindi - English. We also make our code, methodology and approach public to the research
community. 4
2. Literature Review
Discerning hateful content on social media is an already tricky problem given the challenges
associated with it, for instance disrespectful/abusive words could be censored in text, some
expressions may not be inherently oensive, however they can be so in the right context [
9
].
Owing to the conversational design of social media wherein users can reply to a given comment
(either support, refute or irrelevant to the original message), the build-up of threads in response
to a hateful message can also intensify hate even if the reply is not hateful on its own. The
evolution of such hate intensity has shown diverse patterns and no direct correlation to the
parent tweet which makes the task of hate speech detection more dicult [10].
Signicant amount of research has been conducted to evaluate traditional NLP approaches
such as character level CNNs, word embedding based approaches and the myriad of varia-
tions with LSTMs (sub-word level, hierarchical, BiLSTMs) [
11
]. Likewise, Machine Learning
algorithms including SVMs, K-Nearest Neighbours, Multinomial Naive Bayes (MNB) and their
respective performances in multilingual text settings have also been explored [
12
,
13
,
14
]. Investi-
gating categories of profane words that are commonly used in hate speech is another non-trivial
sub-task under the hate detection umbrella, primarily because of the dierent interpretations of
words in dierent cultures/demographics, adaptation of slangs in newer generations etc [15].
In recent times however, with the introduction of Transformer based models and their
performance in Natural Language Understanding (NLU) tasks, signicant work has been done
in order to adapt these for multilingual texts as well to leverage transfer between languages.
Models such as XLMR, mBERT, MuRIL, RemBERT have gained much popularity and have shown
promising results [
16
,
17
,
18
]. Transfer learning based approaches that leverage performance of
high resource languages accompanied with CNN classication heads have also shown signicant
4https://github.com/Adi2K/Precog-HASOC-2021
improvements in capturing hateful content on social media platforms [
19
,
20
]. Sharing and
re-utilizing the model weights learnt whilst training on a corpus for a high resource language
can aid the process of training for languages that are still under explored [21].
3. Dataset
3.1. Dataset & Task Description
Subtask 1
consisted of data for 3 languages, namely – English, Hindi and Marathi [
22
,
23
].
For English and Hindi, the task was further subdivided into 2 sub-parts:
a)
Identication of
hateful v/s non-hateful content and
b)
Characterizing the kind of hate present in a tweet –
either Profane, Hateful, Oensive or None. The distribution of the dierent data classes for
each of the three languages is shown in Table 1.
Language
Number of Tweets
Task A Task B
Non-Hateful Hateful None Oensive Hate Profane
English 1342 2501 1342 1196 683 622
Hindi 3161 1433 3161 654 566 213
Marathi 1205 669 - - - -
Sub Task 2 2899 2841 - - - -
Table 1
Distribution of the HASOC 2021 dataset for Subtask 1, for the three languages. For each language and
task, the corresponding number of tweets per class is shown above.
The focus of
Sub task 2
was binary classication : Hate & Oensive or Non Hate-Oensive.
The given dataset was accompanied by the following additions [24]:
Tweets are English - Hindi Code Mix sentences, and
Classication should not be based on the tweet alone, but should also account for the
context as well.
For example : Consider that in a tweet thread, tweet A is a reply to tweet B. For classifying
tweet A, the model can leverage the information from the parent tweet - tweet B.
Figure 3b demonstrates the relationship between the tweets to be classied and their contexts.
3.2. Preprocessing Data
As a precursor to applying any NLP models on text data, we pre-processed the dataset with
standard techniques. Given that the data fom Twitter is bound to have certain amount of noise
and unwanted elements such as – URLs, mentions etc, these were removed from the tweet texts.
Hashtags have a slightly dierent contribution to analysis of the tweet since they may or may
not contribute positively in the classication task. Through the results from our experiments,
we observed that omitting the hashtags proved to work better, and hence they were cleaned
from the tweet as well.
(a)
Using BerTweet model with a MLP classier
head.
(b)
Combining CNN features over XLM-R output
and manually generated feature vectors.
Figure 1:
Architecture and pipeline for the models used for the downstream task of hate detection and
classification for the English language.
Since the data is code mixed, not only in terms of the combination of languages but also
with respect to scripts (some English text is written in Roman script, whereas some Hindi text
is written in Devanagari apart from Roman), we also normalize the Indic language scripts for
Marathi and Hindi. In addition to that, we removed stop words for the Marathi dataset using
this list. 5Finally, punctuations were also removed from the dataset texts.
An interesting observation was that for the task of hate detection, the presence of emojis
converted to text in the tweets did not improve the performance of our models signicantly
(rather it reduced the scores by some margin). However, including emojis along with text while
classifying hate did have a positive impact since the emoji-text conversion was able to capture
hints of sentiment and indirect oensive/profane content.
4. Methodology
4.1. Sub Task 1: Identifying Hate, oensive and profane content from the
post
4.1.1. English Classifiers
For the English Sub Task, the architecture that resulted in the best performance is an ensemble
of the following models:
Fine-tuned BERTweet model [25]
Fine-tuned XLM-Roberta [26] with CNN Head
5https://github.com/stopwords-iso/stopwords-mr
We use XLM-R, a multilingual model, along with the monolingual model in the ensemble
as we found that some of the text in the training set has transliterated Hindi along with some
Devanagari text. We extracted textual features such as distribution of ‘?’, ‘!’, capital letters etc.
We also use the percentage of profane words and sentiment of the text as a feature. We use
profane words list curated from various sources such as words/cuss
6
, zacanger/profane-words
7
,
t-davison/lexicons.
8
For sentiment analysis we use the TweetEval[
27
] model and use its softmax
output as a feature to our models.
Inspired by Kim [
28
] we pass the embedding (concatenated last 4 hidden layers) to a CNN
and max-pool convolution layers of various widths to a fully connected layer of size 128 with
dropout. We concatenate this 128 dimensional vector with our feature vector. We pass this
output onto a dense output layer with softmax activation and cross entropy loss as shown in
Figure 1.
Along with the previous models, we ne-tune BERTweet, a pre-trained language model for
English tweets. BERTweet has the same architecture as BERT and is trained on the pre-training
procedure of RoBERTa, but it is trained solely on tweets, thus, making it a viable alternative and
suitable for our task. This model has shown state-of-the-art results on tasks based on tweets
[
25
]. We use the encoder architecture and pass the pooled output through a linear layer for the
classication which uses softmax activation and cross-entropy loss as shown in the Figure 1.
We also train the models on the previous years datasets but notice that this does not increase
the performance of the models but actually degrades the performance in Task 1B due to skewed
distribution of classes. Transliteration of emojis didn’t improve the performance. The class
imbalance in task 1B degraded the performance of our models hence we tried to improve upon
it by using a weighted loss function but we notice that this decreases the performance and
that the domain specic distribution is actually helping the models. We also perform K-Fold
Validation and use early stopping to avoid over-tting. We average the probabilities of each
class across folds and the two models in our ensemble.
4.1.2. Hindi & Marathi Classifier
For both the Hindi and Marathi language, the architecture that performed the best utilized the
XLM-R
transformer model. This model was able to capture the code-mixed and multilingual
nature of the tweets dataset. To amplify the results, we leveraged intermediary representations of
the language model as well as textual features that were extracted from the tweets. In particular,
we utilized the Multilingual MiniLM language model for netuning on Hindi Subtask B. We
observed that MiniLM with Focal Loss instead of Cross Entropy Loss performed better than
other baselines in the imbalanced multi-class setting of Hindi Subtask B. Focal Loss compensates
for class imbalance with a factor that increases the network’s sensitivity towards misclassied
samples.
Inspired by Mozafari et al. [
19
] we use the pre-trained representations of the text from 12
hidden layer of XLM-R model (each of 768 dimensions) and then apply a CNN layer with a
kernel size of 3. The output is then passed through a soft-max following which the cross-entropy
6https://github.com/words/cuss
7https://github.com/zacanger/profane-words
8https://github.com/t-davidson/hate-speech-and- oensive-language/tree/master/lexicons
(a)
The base architecture for Hindi & Marathi sub-
tasks using XLM-R with CNN augmented with
textual features vector followed by a softmax
layer.
(b)
Multilingual MiniLM architechture adopted to
overcome class imbalance while characterizing
hate for the Hindi subtask.
Figure 2:
Architecture and pipeline for the models used for the downstream task of hate detection and
classification for the Hindi & Marathi language.
loss is computed whilst training. This model architecture is represented in Figure 2. Tuning
hyperparameters such as optimizers, loss functions and dropout layers, we experiment with
dierent options. For the optimizers we try Adadelta and Adam optimizers with Adam working
out better. Amongst all loss functions, the Cross Entropy Loss performed the best. As for the
dropout layers we explore dropouts in the range 0.1-0.5 and use 0.5 as the nal dropout for the
model architecture.
We further augment the model features, with two kinds of textual features – fraction of
profane words and sentiment of the tweet. Due to lack of resources for Marathi we catalogue
9
a list of profane words in Marathi and use this to nd the fraction of profane words in a tweet.
For Hindi, we curate a list of profane words by collating and appending to existing lists
10
,
and use this to score each tweet. As for the sentiment of the tweet, we incorporated o-the-
shelf HuggingFace models to obtain the positive, negative and neutral scores for a tweet
11 12
.
Although the textual features improved the performance for Hindi only by a small margin, for
Marathi, manually extracted textual features helped in achieving a signicant boost.
For the Marathi Sub Task, we experimented with a voting ensemble of the
XLM-Roberta
with CNN Head using the following features:
Word Embedding + Fraction of Profane Words + Sentiment Polarity
9https://github.com/Adi2K/MarathiSwear
10
https://github.com/neerajvashistha/online-hate-speech-recog/blob/master/data/hi/
Hinglish-Oensive-Text-Classication/Hinglish_Profanity_List.csv
11https://huggingface.co/l3cube-pune/MarathiSentiment
12https://huggingface.co/cardinlp/twitter-xlm-roberta-base-sentiment
(a)
Model pipeline for hate detection in conversa-
tional threads for Sub task 2.
(b)
Hierarchy of a conversation thread and its as-
sociated comments.
Figure 3: Model Pipeline and tweet conversation thread example for sub task 2.
Word Embedding + Sentiment Polarity
Word Embedding
However we noticed that the base model with the embedding and the textual features performed
better on the leaderboard.
4.2. Sub Task 2 : Identification of Conversational Hate-Speech in
Code-Mixed Languages (ICHCL)
The tweets for Sub task 2 are code mixed. While the Transformer based encoder models have
performed well on various monolingual NLU tasks, their performance does not reach the same
level on code mixed sentences. Multilingual transformer based models, have been applied
for various code mixed NLU tasks, and have performed better than monolingual transformer
based models [
29
]. For this task, we use XLM-RoBERTa [
26
]. To capture the context and the
tweet itself, we modify the input in the following manner, where [CLS] , [SEP] are part of the
vocabulary of model, and are used to classify and take multiple sentences as input, respectively.
[CLS] <Tweet text to be classied> [SEP] <context of parent tweet> [SEP]
Here, <Tweet text to be classied> is the text of the tweet/comment/reply that is being classied,
while <Context of parent tweet> is either just the parent tweet or concatenation of parent tweet
and comment, depending on weather the text to be classied is a tweet or a comment or a reply.
While classifying a standalone tweet, the context is left empty. The Hindi corpus used to train
XLM-Roberta is in Devanagari script, while there is only a small portion of the corpus which
is in Romanised form. With the hypothesis that the performance of model will improve if the
Hindi tokens are in Devanagari script, we used CSNLI tool
13
to convert the Romanised tokens
to Devanagari script. However, this normalisation only had a marginal impact on the nal
performance of the model. We used Huggingface’s Trainer API to train the XLM-R model, and
the hyperparameters were choosen using the hyperparameter search functionality oered by
Trainer API.
4.3. Experiments
We used Huggingface Transformers [
30
] library for implementing the classiers. For hyper
parameter tuning we use Optuna Framework
14
library. Exploring multiple architectures simul-
taneously, we also tried ensembling an odd number of models following a majority rule based
selection. For the English subtask we also did ensembling with averaged softmax probabilities.
However, the increase in complexity of the classication pipeline did not necessarily improve
performance scores, considering the size and the distribution of the dataset for Hindi and
Marathi but helped in English. Table 2captures the Accuracies and F1 scores (corresponding to
submissions made on the leaderboard) of all our models for each of the sub tasks.
Language Sub Task Method Accuracy Macro F1
English A
XLM-R + CNN 62.30 0.62
Ensemble 79.94 0.78
XLM-R + CNN + Sentiment Scores 81.03 0.79
English B XLM-R + CNN + Weighted loss 60.50 0.53
Ensemble 65.18 0.59
Hindi A
MuRIL 68.90 0.60
XLM-R Base 74.00 0.76
XLM-R + CNN 80.09 0.77
Hindi B MiniLM with Focal Loss 72.64 0.51
Marathi A
XLM-R Base 84.16 0.86
Ensemble 88.48 0.87
XLM-R + CNN 88.64 0.87
Hi-En Code mix 2 XLM-R without norm 67.58 0.67
XLM-R with norm 69.36 0.70
Table 2
We can see performance scores for each of the six subtasks in terms of their test accuracy percentage
and Macro F1 scores. All the architectures that were experimented and tested out are tabulated here.
We can observe form the results that XLM-R combined with CNN classifier head works best across the
languages of Sub Task 1, while for Subtask 2, XLM-R with normalised input text performs the best in
our experiments.
13https://github.com/irshadbhat/csnli
14https://optuna.org/
5. Conclusion
In this paper, we presented our approaches for Hate Speech detection on Indian Languages
and code mix between Hindi-English using multilingual transformer based encoder models.
Although, in this work we have employed dierent models to address individual language
specic sub-tasks, a multi-task single model based approach, which performs well across all
the language pairs, would be an interesting challenge, which we wish to explore as a future
work. In addition to this, as part of future work, we would like to improve the performance by
carrying out an additional step of domain adaptive pre-training of the encoder models, and an
ecient ensemble of multilingual encoder models.
Acknowledgments
We would like to thank the organisers of HASOC’21 Shared task for addressing a crucial problem
of hate speech in Indian languages by releasing data resources, and for the smooth conduct of
the competition. We would also like to specially thank all members of our research lab, Precog,
for the constructive suggestions during the whole process.
References
[1]
B. Mathew, R. Dutt, P. Goyal, A. Mukherjee, Spread of hate speech in online social media,
in: Proceedings of the 10th ACM Conference on Web Science, WebSci ’19, Association
for Computing Machinery, New York, NY, USA, 2019, p. 173–182. URL: https://doi.org/10.
1145/3292522.3326034. doi:10.1145/3292522.3326034.
[2]
L. Silva, M. Mondal, D. Correa, F. Benevenuto, I. Weber, Analyzing the targets of hate in
online social media, in: Tenth international AAAI conference on web and social media,
2016.
[3]
M. Mozafari, R. Farahbakhsh, N. Crespi, Hate speech detection and racial bias mitigation
in social media based on bert model, PLOS ONE 15 (2020) 1–26. URL: https://doi.org/10.
1371/journal.pone.0237861. doi:10.1371/journal.pone.0237861.
[4] Z. Mossie, J.-H. Wang, Vulnerable community identication using hate speech detection
on social media, Information Processing & Management 57 (2020) 102087. URL: https:
//www.sciencedirect.com/science/article/pii/S0306457318310902. doi:
https://doi.org/
10.1016/j.ipm.2019.102087.
[5]
K. A. Qureshi, M. Sabih, Un-compromised credibility: Social media based multi-class hate
speech classication for text, IEEE Access 9 (2021) 109465–109477. doi:
10.1109/ACCESS.
2021.3101977.
[6]
K. Saha, E. Chandrasekharan, M. De Choudhury, Prevalence and psychological eects of
hateful speech in online college communities, in: Proceedings of the 10th ACM Conference
on Web Science, WebSci ’19, Association for Computing Machinery, New York, NY, USA,
2019, p. 255–264. URL: https://doi.org/10.1145/3292522.3326032. doi:
10.1145/3292522.
3326032.
[7]
T. Ranasinghe, M. Zampieri, An evaluation of multilingual oensive language identication
methods for the languages of india, Information 12 (2021). URL: https://www.mdpi.com/
2078-2489/12/8/306. doi:10.3390/info12080306.
[8]
b. . F. Modha, Sandip and Mandl, Thomas and Shahi, Gautam Kishore and Madhu, Hiren
and Satapara, Shrey and Ranasinghe, Tharindu and Zampieri, Marcos, Overview of the
HASOC Subtrack at FIRE 2021: Hate Speech and Oensive Content Identication in
English and Indo-Aryan Languages and Conversational Hate Speech, ACM, 2021.
[9]
G. Kovács, P. Alonso, R. Saini, Challenges of hate speech detection in social media, SN
Computer Science 2 (2021) 95. URL: https://doi.org/10.1007/s42979-021-00457-3. doi:
10.
1007/s42979-021-00457-3.
[10]
S. Dahiya, S. Sharma, D. Sahnan, V. Goel, E. Chouzenoux, V. Elvira, A. Majumdar, A. Band-
hakavi, T. Chakraborty, Would your tweet invoke hate on the y? forecasting hate intensity
of reply threads on twitter, in: Proceedings of the 27th ACM SIGKDD Conference on
Knowledge Discovery & Data Mining, KDD ’21, Association for Computing Machinery,
New York, NY, USA, 2021, p. 2732–2742. URL: https://doi.org/10.1145/3447548.3467150.
doi:10.1145/3447548.3467150.
[11]
T. Y. Santosh, K. V. Aravind, Hate speech detection in hindi-english code-mixed social media
text, in: Proceedings of the ACM India Joint International Conference on Data Science and
Management of Data, CoDS-COMAD ’19, Association for Computing Machinery, New
York, NY, USA, 2019, p. 310–313. URL: https://doi.org/10.1145/3297001.3297048. doi:
10.
1145/3297001.3297048.
[12]
P. Rani, S. Suryawanshi, K. Goswami, B. R. Chakravarthi, T. Fransen, J. P. McCrae, A
comparative study of dierent state-of-the-art hate speech detection methods in Hindi-
English code-mixed data, in: Proceedings of the Second Workshop on Trolling, Aggression
and Cyberbullying, European Language Resources Association (ELRA), Marseille, France,
2020, pp. 42–48. URL: https://aclanthology.org/2020.trac-1.7.
[13]
T. Ranasinghe, M. Zampieri, An evaluation of multilingual oensive language identication
methods for the languages of india, Information 12 (2021). URL: https://www.mdpi.com/
2078-2489/12/8/306. doi:10.3390/info12080306.
[14]
F. E. Ayo, O. Folorunso, F. T. Ibharalu, I. A. Osinuga, Machine learning tech-
niques for hate speech classication of twitter data: State-of-the-art, future challenges
and research directions, Computer Science Review 38 (2020) 100311. URL: https://
www.sciencedirect.com/science/article/pii/S1574013720304111. doi:
https://doi.org/
10.1016/j.cosrev.2020.100311.
[15]
P. L. Teh, C.-B. Cheng, W. M. Chee, Identifying and categorising profane words in hate
speech, in: Proceedings of the 2nd International Conference on Compute and Data
Analysis, ICCDA 2018, Association for Computing Machinery, New York, NY, USA, 2018, p.
65–69. URL: https://doi.org/10.1145/3193077.3193078. doi:10.1145/3193077.3193078.
[16]
J. Devlin, M.-W. Chang, K. Lee, K. Toutanova, BERT: Pre-training of deep bidirectional
transformers for language understanding, in: Proceedings of the 2019 Conference of
the North American Chapter of the Association for Computational Linguistics: Human
Language Technologies, Volume 1 (Long and Short Papers), Association for Computational
Linguistics, Minneapolis, Minnesota, 2019, pp. 4171–4186. URL: https://aclanthology.org/
N19-1423. doi:10.18653/v1/N19-1423.
[17]
A. Conneau, K. Khandelwal, N. Goyal, V. Chaudhary, G. Wenzek, F. Guzmán, E. Grave,
M. Ott, L. Zettlemoyer, V. Stoyanov, Unsupervised cross-lingual representation learning at
scale, in: Proceedings of the 58th Annual Meeting of the Association for Computational
Linguistics, Association for Computational Linguistics, Online, 2020, pp. 8440–8451. URL:
https://aclanthology.org/2020.acl-main.747. doi:10.18653/v1/2020.acl-main.747.
[18]
S. Khanuja, D. Bansal, S. Mehtani, S. Khosla, A. Dey, B. Gopalan, D. K. Margam, P. Aggarwal,
R. T. Nagipogu, S. Dave, S. Gupta, S. C. B. Gali, V. Subramanian, P. Talukdar, Muril:
Multilingual representations for indian languages, 2021. arXiv:2103.10730.
[19]
M. Mozafari, R. Farahbakhsh, N. Crespi, A bert-based transfer learning approach for hate
speech detection in online social media, in: H. Cheri, S. Gaito, J. F. Mendes, E. Moro,
L. M. Rocha (Eds.), Complex Networks and Their Applications VIII, Springer International
Publishing, Cham, 2020, pp. 928–940.
[20]
I. Bigoulaeva, V. Hangya, A. Fraser, Cross-lingual transfer learning for hate speech
detection, in: Proceedings of the First Workshop on Language Technology for Equality,
Diversity and Inclusion, Association for Computational Linguistics, Kyiv, 2021, pp. 15–25.
URL: https://aclanthology.org/2021.ltedi-1.3.
[21]
T. Ranasinghe, M. Zampieri, Multilingual oensive language identication for low-
resource languages, CoRR abs/2105.05996 (2021). URL: https://arxiv.org/abs/2105.05996.
arXiv:2105.05996.
[22]
T. Mandl, S. Modha, G. K. Shahi, H. Madhu, S. Satapara, P. Majumder, J. Schäfer, T. Ranas-
inghe, M. Zampieri, D. Nandini, A. K. Jaiswal, Overview of the HASOC subtrack at FIRE
2021: Hate Speech and Oensive Content Identication in English and Indo-Aryan Lan-
guages, in: Working Notes of FIRE 2021 - Forum for Information Retrieval Evaluation,
CEUR, 2021. URL: http://ceur-ws.org/.
[23]
S. Gaikwad, T. Ranasinghe, M. Zampieri, C. M. Homan, Cross-lingual oensive language
identication for low resource languages: The case of marathi, in: Proceedings of RANLP,
2021.
[24]
S. Satapara, S. Modha, T. Mandl, H. Madhu, P. Majumder, Overview of the HASOC
Subtrack at FIRE 2021: Conversational Hate Speech Detection in Code-mixed language ,
in: Working Notes of FIRE 2021 - Forum for Information Retrieval Evaluation, CEUR, 2021.
[25]
D. Q. Nguyen, T. Vu, A. T. Nguyen, BERTweet: A pre-trained language model for English
Tweets, in: Proceedings of the 2020 Conference on Empirical Methods in Natural Language
Processing: System Demonstrations, 2020, pp. 9–14.
[26]
A. Conneau, K. Khandelwal, N. Goyal, V. Chaudhary, G. Wenzek, F. Guzmán, E. Grave,
M. Ott, L. Zettlemoyer, V. Stoyanov, Unsupervised cross-lingual representation learning at
scale, in: Proceedings of the 58th Annual Meeting of the Association for Computational
Linguistics, Association for Computational Linguistics, Online, 2020, pp. 8440–8451. URL:
https://aclanthology.org/2020.acl-main.747. doi:10.18653/v1/2020.acl-main.747.
[27]
F. Barbieri, J. Camacho-Collados, L. Espinosa-Anke, L. Neves, TweetEval:Unied Bench-
mark and Comparative Evaluation for Tweet Classication, in: Proceedings of Findings of
EMNLP, 2020.
[28]
Y. Kim, Convolutional neural networks for sentence classication, in: Proceedings of
the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP),
Association for Computational Linguistics, 2014, pp. 1746–1751. URL: https://aclanthology.
org/D14-1181.
[29]
S. Khanuja, S. Dandapat, A. Srinivasan, S. Sitaram, M. Choudhury, GLUECoS: An evaluation
benchmark for code-switched NLP, in: Proceedings of the 58th Annual Meeting of the
Association for Computational Linguistics, Association for Computational Linguistics,
Online, 2020, pp. 3575–3585. URL: https://aclanthology.org/2020.acl-main.329. doi:
10.
18653/v1/2020.acl-main.329.
[30]
T. Wolf, L. Debut, V. Sanh, J. Chaumond, C. Delangue, A. Moi, P. Cistac, T. Rault, R. Louf,
M. Funtowicz, J. Davison, S. Shleifer, P. von Platen, C. Ma, Y. Jernite, J. Plu, C. Xu, T. Le Scao,
S. Gugger, M. Drame, Q. Lhoest, A. Rush, Transformers: State-of-the-art natural lan-
guage processing, in: Proceedings of the 2020 Conference on Empirical Methods in
Natural Language Processing: System Demonstrations, Association for Computational
Linguistics, Online, 2020, pp. 38–45. URL: https://aclanthology.org/2020.emnlp-demos.6.
doi:10.18653/v1/2020.emnlp-demos.6.
ResearchGate has not been able to resolve any citations for this publication.
Conference Paper
Full-text available
The widespread presence of offensive language on social media motivated the development of systems capable of recognizing such content automatically. Apart from a few notable exceptions, most research on automatic offensive language identification has dealt with English. To address this shortcoming , we introduce MOLD, the Marathi Offensive Language Dataset. MOLD is the first dataset of its kind compiled for Marathi, thus opening a new domain for research in low-resource Indo-Aryan languages. We present results from several machine learning experiments on this dataset, including zero-short and other transfer learning experiments on state-of-the-art cross-lingual transformers from existing data in Bengali, English, and Hindi.
Article
Full-text available
There is an enormous growth of social media which fully promotes freedom of expression through its anonymity feature. Freedom of expression is a human right but hate speech towards a person or group based on race, caste, religion, ethnic or national origin, sex, disability, gender identity, etc. is an abuse of this sovereignty. It seriously promotes violence or hate crimes and creates an imbalance in society by damaging peace, credibility, and human rights, etc. Detecting hate speech in social media discourse is quite essential but a complex task. There are different challenges related to appropriate and social media-specific dataset availability and its high-performing supervised classifier for text-based hate speech detection. These issues are addressed in this study, which includes the availability of social media-specific broad and balanced dataset, with multi-class labels and its respective automatic classifier, a dataset with language subtleties, dataset labeled under a comprehensive definition and well-defined rules, dataset labeled with the strong agreement of annotators, etc. Addressing different categories of hate separately, this paper aims to accurately predict their different forms, by exploring a group of text mining features. Two distinct groups of features are explored for problem suitability. These are baseline features and self-discovered/new features. Baseline features include the most commonly used effective features of related studies. Exploration found a few of them, like character and word n-grams, dependency tuples, sentiment scores, and count of 1st, 2nd person pronouns are more efficient than others. Due to the application of latent semantic analysis (LSA) for dimensionality reduction, this problem is benefited from the utilization of many complex and non-linear models and CAT Boost performed best. The proposed model is compared with related studies in addition to system baseline models. The results produced by the proposed model were much appreciating.
Conference Paper
Full-text available
Curbing hate speech is undoubtedly a major challenge for online microblogging platforms like Twitter. While there have been studies around hate speech detection, it is not clear how hate speech finds its way into an online discussion. It is important for a content moderator to not only identify which tweet is hateful, but also to predict which tweet will be responsible for accumulating hate speech. This would help in prioritizing tweets that need constant monitoring. Our analysis reveals that for hate speech to manifest in an ongoing discussion, the source tweet may not necessarily be {\em hateful}; rather, there are plenty of such non-hateful tweets which gradually invoke hateful replies, resulting in the entire reply threads becoming provocative. In this paper, we define a novel problem -- {\em given a source tweet and a few of its initial replies, the task is to forecast the hate intensity of upcoming replies}. To this end, we curate a novel dataset constituting $\sim 4.5k$ contemporary tweets and their entire reply threads. Our preliminary analysis confirms that the evolution patterns along time of hate intensity among reply threads have highly diverse patterns, and there is no significant correlation between the hate intensity of the source tweets and that of their reply threads. We employ seven state-of-the-art dynamic models (either statistical signal processing or deep learning based) and show that they fail badly to forecast the hate intensity. We then propose \name, a novel deep state-space model that leverages the function approximation capability of deep neural networks with the capacity to quantify the uncertainty of statistical signal processing models. Exhaustive experiments and ablation study show that \name\ outperforms all the baselines substantially. Further, its deployment in an advanced AI platform designed to monitor real-world problematic hateful content has improved the aggregated insights extracted for countering the spread of online harms.
Article
Full-text available
The detection of hate speech in social media is a crucial task. The uncontrolled spread of hate has the potential to gravely damage our society, and severely harm marginalized people or groups. A major arena for spreading hate speech online is social media. This significantly contributes to the difficulty of automatic detection, as social media posts include paralinguistic signals (e.g. emoticons, and hashtags), and their linguistic content contains plenty of poorly written text. Another difficulty is presented by the context-dependent nature of the task, and the lack of consensus on what constitutes as hate speech, which makes the task difficult even for humans. This makes the task of creating large labeled corpora difficult, and resource consuming. The problem posed by ungrammatical text has been largely mitigated by the recent emergence of deep neural network (DNN) architectures that have the capacity to efficiently learn various features. For this reason, we proposed a deep natural language processing (NLP) model—combining convolutional and recurrent layers—for the automatic detection of hate speech in social media data. We have applied our model on the HASOC2019 corpus, and attained a macro F1 score of 0.63 in hate speech detection on the test set of HASOC. The capacity of DNNs for efficient learning, however, also means an increased risk of overfitting. Particularly, with limited training data available (as was the case for HASOC). For this reason, we investigated different methods for expanding resources used. We have explored various opportunities, such as leveraging unlabeled data, similarly labeled corpora, as well as the use of novel models. Our results showed that by doing so, it was possible to significantly increase the classification score attained.
Article
Full-text available
Disparate biases associated with datasets and trained classifiers in hateful and abusive content identification tasks have raised many concerns recently. Although the problem of biased datasets on abusive language detection has been addressed more frequently, biases arising from trained classifiers have not yet been a matter of concern. In this paper, we first introduce a transfer learning approach for hate speech detection based on an existing pre-trained language model called BERT (Bidirectional Encoder Representations from Transformers) and evaluate the proposed model on two publicly available datasets that have been annotated for racism, sexism, hate or offensive content on Twitter. Next, we introduce a bias alleviation mechanism to mitigate the effect of bias in training set during the fine-tuning of our pre-trained BERT-based model for hate speech detection. Toward that end, we use an existing regularization method to reweight input samples, thereby decreasing the effects of high correlated training set’ s n-grams with class labels, and then fine-tune our pre-trained BERT-based model with the new re-weighted samples. To evaluate our bias alleviation mechanism, we employed a cross-domain approach in which we use the trained classifiers on the aforementioned datasets to predict the labels of two new datasets from Twitter, AAE-aligned and White-aligned groups, which indicate tweets written in African-American English (AAE) and Standard American English (SAE), respectively. The results show the existence of systematic racial bias in trained classifiers, as they tend to assign tweets written in AAE from AAE-aligned group to negative classes such as racism, sexism, hate, and offensive more often than tweets written in SAE from White-aligned group. However, the racial bias in our classifiers reduces significantly after our bias alleviation mechanism is incorporated. This work could institute the first step towards debiasing hate speech and abusive language detection systems.
Article
Twitter is a microblogging tool that allow the creation of big data through short digital contents. This study provides a survey of machine learning techniques for hate speech classification from Twitter data streams. Hate speech classification in Twitter data streams has remain a vibrant research focus, but little research efforts have been devoted to the design of a generic metadata architecture, threshold settings and fragmentation issues. Hate speech classification techniques presented in literature address some of the challenges inherent in Twitter data streams but limited in the aforementioned issues. This study presented collection of hate speech benchmarks datasets suitable for testing the efficiency of classification models. This study also presented the pros and cons for single and hybrid machine learning methods in hate speech classification. The summary of performance evaluation for the surveyed machine learning methods was also presented. The study also presented a generic metadata architecture for hate speech classification in Twitter to tackle issues with Twitter data streams. The developed generic metadata architecture was observed to performed better across all evaluation metrics for hate speech detection having 0.95, 0.93, 0.92 and 0.93 for accuracy, precision, recall and F1-score respectively, when compared to similar methods. Similarly, the developed generic metadata architecture for hate speech sentiment classification performed better with F1-score of 91.5% compared to related methods. The developed generic metadata architecture also indicates a more perfect test having an AUC of 0.97, when compared to similar methods. The statistical validation of results points out the efficiency of the developed system. Finally, the results also showed that the developed system is very good for automatic topic detection and categorization.