Social media platforms are used by a large number of people prominently to express their thoughts and opinions. However, these platforms have contributed to a sub stantial amount of hateful and abusive content as well. Therefore, it is impor tant to curb the spread of hate speech on these platforms. In India, Marathi is one of the most popular languages used by a wide audience. In this work, we present L3Cube-MahaHate, the first ma jor Hate Speech Dataset in Marathi. The dataset is curated from Twitter, anno tated manually. Our dataset consists of over 00 distinct tweets labeled into four major classes i.e hate, offensive, pro fane, and not. We present the approaches used for collecting and annotating the data and the challenges faced during the pro cess. Finally, we present baseline classi fication results using deep learning mod els based on CNN, LSTM, and Transform ers. We explore mono-lingual and multi lingual variants of BERT like MahaBERT, IndicBERT, mBERT, and xlm-RoBERTa and show that mono-lingual models per form better than their multi-lingual coun terparts. The MahaBERT model provides the best results on L3Cube-MahaHate Corpus.

The proliferation of online hate speech has necessitated the creation of algorithms which can detect toxicity. Most of the past research focuses on this detection as a classification task, but assigning an absolute toxicity label is often tricky. Hence, few of the past works transform the same task into a regression. This paper shows the comparative evaluation of different transformers and traditional machine learning models on a recently released toxicity severity measurement dataset by Jigsaw. We further demonstrate the issues with the model predictions using explainability analysis.

Automated textual cyberbullying detection is known to be a challenging task. It is sometimes expected that messages associated with bullying will either be a) abusive, b) targeted at a specific individual or group, or c) have a negative sentiment. Transfer learning by fine-tuning pre-trained attention-based transformer language models (LMs) has achieved near state-of-the-art (SOA) precision in identifying textual fragments as being bullying-related or not. This study looks closely at two SOA LMs, BERT and HateBERT, fine-tuned on real-life cyberbullying datasets from multiple social networking platforms. We intend to determine whether these finely calibrated pre-trained LMs learn textual cyberbullying attributes or syntactical features in the text. The results of our comprehensive experiments show that despite the fact that attention weights are drawn more strongly to syntactical features of the text at every layer, attention weights cannot completely account for the decision-making of such attention-based transformers.

In this paper, we presented our team "IIITRanchi” for the Trolling, Aggression and Cyberbullying (TRAC-3) 2022 shared tasks. Aggression and its different forms on social media and other platforms had tremendous growth on the Internet. In this work we have tried upon different aspects of aggression, aggression intensity, bias of different forms and their usage online and its identification using different Machine Learning techniques. We have classified each sample at seven different tasks namely aggression level, aggression intensity, discursive role, gender bias, religious bias, caste/class bias and ethnicity/racial bias as specified in the shared tasks. Both of our teams tried machine learning classifiers and achieved the good results. Overall, our team "IIITRanchi” ranked first position in this shared tasks competition.

Online hate speech detection is an inherently challenging task that has recently received much attention from the natural language processing community. Despite a substantial increase in performance, considerable challenges remain and include encoding contextual information into automated hate speech detection systems. In this paper, we focus on detecting the target of hate speech in Dutch social media: whether a hateful Facebook comment is directed against migrants or not (i.e., against someone else). We manually annotate the relevant conversational context and investigate the effect of different aspects of context on performance when adding it to a Dutch transformer-based pre-trained language model, BERTje. We show that performance of the model can be significantly improved by integrating relevant contextual information.

In this paper, we discuss an interpretable framework to integrate toxic language annotations. Most data sets address only one aspect of the complex relationship in toxic communication and are inconsistent with each other. Enriching annotations with more details and information is however of great importance in order to develop high-performing and comprehensive explainable language models. Such systems should recognize and interpret both expressions that are toxic as well as expressions that make reference to specific targets to combat toxic language. We therefore created a crowd-annotation task to mark the spans of words that refer to target communities as an extension of the HateXplain data set. We present a quantitative and qualitative analysis of the annotations. We also fine-tuned RoBERTa-base on our data and experimented with different data thresholds to measure their effect on the classification. The F1-score of our best model on the test set is 79%. The annotations are freely available and can be combined with the existing HateXplain annotation to build richer and more complete models.

Annotating abusive language is expensive, logistically complex and creates a risk of psychological harm. However, most machine learning research has prioritized maximizing effectiveness (i.e., F1 or accuracy score) rather than data efficiency (i.e., minimizing the amount of data that is annotated). In this paper, we use simulated experiments over two datasets at varying percentages of abuse to demonstrate that transformers-based active learning is a promising approach to substantially raise efficiency whilst still maintaining high effectiveness, especially when abusive content is a smaller percentage of the dataset. This approach requires a fraction of labeled data to reach performance equivalent to training over the full dataset.

Highly imbalanced textual datasets continue to pose a challenge for supervised learning models. However, viewing such imbalanced text data as an anomaly detection (AD) problem has advantages for certain tasks such as detecting hate speech, or inappropriate and/or offensive language in large social media feeds. There the unwanted content tends to be both rare and non-uniform with respect to its thematic character, and better fits the definition of an anomaly than a class. Several recent approaches to textual AD use transformer models, achieving good results but with trade-offs in pre-training and inflexibility with respect to new domains. In this paper we compare two linear models within the NMF family, which also have a recent history in textual AD. We introduce a new approach based on an alternative regularization of the NMF objective. Our results surpass other linear AD models and are on par with deep models, performing comparably well even in very small outlier concentrations.

Political competitions are complex settings where candidates use campaigns to promote their chances to be elected. One choice focuses on conducting a positive campaign that highlights the candidate’s achievements, leadership skills, and future programs. The alternative is to focus on a negative campaign that emphasizes the negative aspects of the competing person and is aimed at offending opponents or the opponent’s supporters. In this proposal, we concentrate on negative campaigns in Israeli elections. This work introduces an empirical case study on automatic detection of negative campaigns, using machine learning and natural language processing approaches, applied to the Hebrew-language data from Israeli municipal elections. Our contribution is multi-fold: (1) We provide TONIC—daTaset fOr Negative polItical Campaign in Hebrew—which consists of annotated posts from Facebook related to Israeli municipal elections; (2) We introduce results of a case study, that explored several research questions. RQ1: Which classifier and representation perform best for this task? We employed several traditional classifiers which are known for their good performance in IR tasks and two pre-trained models based on BERT architecture; several standard representations were employed with traditional ML models. RQ2: Does a negative campaign always contain offensive language? Can a model, trained to detect offensive language, also detect negative campaigns? We are trying to answer this question by reporting results for the transfer learning from a dataset annotated with offensive language to our dataset.

Standard approaches to hate speech detection rely on sufficient available hate speech annotations. Extending previous work that repurposes natural language inference (NLI) models for zero-shot text classification, we propose a simple approach that combines multiple hypotheses to improve English NLI-based zero-shot hate speech detection. We first conduct an error analysis for vanilla NLI-based zero-shot hate speech detection and then develop four strategies based on this analysis. The strategies use multiple hypotheses to predict various aspects of an input text and combine these predictions into a final verdict. We find that the zero-shot baseline used for the initial error analysis already outperforms commercial systems and fine-tuned BERT-based hate speech detection models on HateCheck. The combination of the proposed strategies further increases the zero-shot accuracy of 79.4% on HateCheck by 7.9 percentage points (pp), and the accuracy of 69.6% on ETHOS by 10.0pp.