Search

Exponential growth in social media has led to the increasing popularity of hate speech and hate-based propaganda. Hate speech or malicious expression refers to the use of offensive, violent, or offensive language and its religious conduct with a specific group of people who share a common property, such as gender, ethnicity, race, or beliefs. Online hate diffusion has now become a serious problem as it creates a series of international initiatives aimed at defining problems and developing effective countermeasures; this study delves into the exploration of the intention of hate speech posting on social media, especially on Twitter. Both dramaturgical models of social interaction and cultivation theory were used to explain the hate speech culture phenomenon. A qualitative method is proposed for this study as part of the exploration. Results revealed that most of the previous studies on hate speech focused on the field of computer science but rarely on the communication field. The paper presents the results of past studies and shows the new proposed framework. The investigation suggests future directions for the problem and possible solutions; it starts with the background of the research, the explanation of the problem, the meaning of the research, and pursuing the research questions and goals of the research before finally explaining the limits. [ABSTRACT FROM AUTHOR]

The internet has revolutionised our ability to communicate and connect across historic social, political and geographic divides. Where previously gatekeepers mitigated and negotiated access to mass media platforms, today potentially anyone – and any content – can reach millions of users in an instant. This development bears great opportunities for the democratisation of expression and the diversification of public discourse but has likewise broadened the impact of harm caused online. This raises the question how platforms and services can be regulated effectively to combat online harms without jeopardising free and open discourse. The paper explores the Online Harms White Paper published by the UK Government earlier this year and compares its regulatory approach with the infamous German Network Enforcement Law. [ABSTRACT FROM AUTHOR]

There has been surge in the usage of Internet as well as social media platforms which has led to rise in online hate speech targeted on individual or group. In the recent years, hate speech has resulted in one of the challenging problems that can unfurl at a fast pace on digital platforms leading to various issues such as prejudice, violence and even genocide. Considering the acceptance of Artificial Intelligence (AI) and Natural Language Processing (NLP) techniques in varied application domains, it would be intriguing to consider these techniques for automated hate speech detection. In literature, there have been efforts to recognize and categorize hate speech using varied Machine Learning (ML) and Deep Learning (DL) techniques. Hence, considering the need and provocations for hate speech detection we aim to present a comprehensive review that discusses fundamental taxonomy as well as recent advances in the field of online hate speech identification. There is a significant amount of literature related to the initial phases of hate speech detection. The background section provides a detailed explanation of the previous research. The subsequent section that follows is dedicated to examining the recent literature published from the year 2020 onwards. The paper presents some of the hate speech datasets considered for hate speech detection. Furthermore, the paper discusses different data modalities, namely, textual hate speech detection, multi‐modal hate speech detection and multilingual hate speech detection. Apart from systematic review on hate speech detection, the paper also implement several multi‐label models to compare the performance of hate speech detection by employing classic ML technique namely, Logistic Regression and DL technique namely, Long Short‐Term Memory (LSTM) and a multiclass multi‐label architecture. In the implemented architecture, we have derived two new elements to quantify the hatefulness and intensity of hatred to improve the results for hate speech detection using Indonesian tweet dataset. Empirical Analysis of the model reveals that the implemented approach outperforms and is able to achieve improved results for the underlying dataset. [ABSTRACT FROM AUTHOR]

The detection of toxic language in the Arabic language has emerged as an active area of research in recent years, and reviewing the existing datasets employed for training the developed solutions has become a pressing need. This paper offers a comprehensive survey of Arabic datasets focused on online toxic language. We systematically gathered a total of 54 available datasets and their corresponding papers and conducted a thorough analysis, considering 18 criteria across four primary dimensions: availability details, content, annotation process, and reusability. This analysis enabled us to identify existing gaps and make recommendations for future research works. For the convenience of the research community, the list of the analysed datasets is maintained in a GitHub repository. [ABSTRACT FROM AUTHOR]

Information and communication technology has evolved dramatically, and now the majority of people are using internet and sharing their opinion more openly, which has led to the creation, collection and circulation of hate speech over multiple platforms. The anonymity and movability given by these social media platforms allow people to hide themselves behind a screen and spread the hate effortlessly. Online hate speech (OHS) recognition can play a vital role in stopping such activities and can thus restore the position of public platforms as the open marketplace of ideas. To study hate speech detection in social media, we surveyed the related available datasets on the web-based platform. We further analyzed approximately 200 research papers indexed in the different journals from 2010 to 2022. The papers were divided into various sections and approaches used in OHS detection, i.e., feature selection, traditional machine learning (ML) and deep learning (DL). Based on the selected 111 papers, we found that 44 articles used traditional ML and 35 used DL-based approaches. We concluded that most authors used SVM, Naive Bayes, Decision Tree in ML and CNN, LSTM in the DL approach. This survey contributes by providing a systematic approach to help researchers identify a new research direction in online hate speech. [ABSTRACT FROM AUTHOR]

In today's digital landscape, organizations face significant challenges, including sensitive data leaks and the proliferation of hate speech, both of which can lead to severe consequences such as financial losses, reputational damage, and psychological impacts on employees. This work considers a comprehensive solution using a microservices architecture to monitor computer usage within organizations effectively. The approach incorporates spyware techniques to capture data from employee computers and a web application for alert management. The system detects data leaks, suspicious behaviors, and hate speech through efficient data capture and predictive modeling. Therefore, this paper presents a comparative performance analysis between Spring Boot and Quarkus, focusing on objective metrics and quantitative statistics. By utilizing recognized tools and benchmarks in the computer science community, the study provides an in-depth understanding of the performance differences between these two platforms. The implementation of Quarkus over Spring Boot demonstrated substantial improvements: memory usage was reduced by up to 80% and CPU usage by 95%, and system uptime decreased by 119%. This solution offers a robust framework for enhancing organizational security and mitigating potential threats through proactive monitoring and predictive analysis while also guiding developers and software architects in making informed technological choices. [ABSTRACT FROM AUTHOR]

In this paper, I argue that considerations of epistemic bubbles can give us reason to defend censorship of hate speech. Although censoring hate speech leads to epistemic bubbles ('censorship bubbles'), they tend to be less epistemically problematic than epistemic bubbles generated by the circulation of hate speech ('hate bubbles'). Because hate speech silences its target groups and creates the illusion that the dominant group identities are threatened, hate bubbles are likely more restrictive in structure than censorship bubbles and have a stronger tendency to turn into echo chambers where opposing views are actively discredited. Therefore, I argue that censorship bubbles might be a minor price we ought to pay to avoid hate bubbles. Additionally, my analysis shows that we cannot focus exclusively on the content or the structure of epistemic bubbles, since their content can partly determine their structure, as is in the case of hate bubbles. [ABSTRACT FROM AUTHOR]

In this paper, we will investigate whether or not the name 'Africa' can be seen as a slur. Since the name was given by European exploiters, slavers and colonists, it raises the question of whether such a name should continue to be accepted or abandoned. It may seem that just as the slavers renamed their victims and the colonists renamed the territories they conquered, the name Africa/ns similarly was an imposition on the continent and its peoples. It can also be argued that the naming of the continent by an external aggressor is a form of epistemic subordination that vitiates the dignity of the inhabitants. That is to say that European slavers took it upon themselves to give the inhabitants an identity that highlights climate and possibly skin colour because it was something they could not do for themselves. This presents the inhabitants as inferior to their namers and whose millennia-old civilisational achievements can comfortably be overlooked. We argue that the two preceding arguments constitute compelling reasons to abandon the name Africa as a compromised identity and offer an idea for the renaming of the continent. [ABSTRACT FROM AUTHOR]

Social media is a great source of communication. People use various social media platforms, such as Twitter, Facebook, and Instagram, for sharing their ideas, opinions, and feelings. Users of different age groups, cultures, education backgrounds manipulate these powerful mediums of communication. Even though it gives all the benefits of knowledge sharing among the users, it has a dark side too. Despite setting restrictions from the corresponding sites, many users use abusive language to blemish the status and image of someone. So it is highly the need of the hour for the government or the particular social media platform to sift out those unwanted hate texts before diffusing them. Finding the hate text is one of the emerging research topics in Natural Language Processing where the model predicts the given text as hate text or not. This automated hate text detection becomes tedious when we consider the Indian languages due to a lack of data. Moreover, Indian people are multilingual and use code-mixed patterns to express their thoughts. The unavailability of the annotated Tamil-English dataset and the lack of a standard model make this task more challenging. In our paper, to handle such code-mixed data, a dataset is created with 10000 Tamil-English code-mixed texts collected from Twitter. These are annotated as hate text/non-hate text. In this paper, we use a synonym-based Bi-LSTM model for classifying hate non-hate text in tweets. [ABSTRACT FROM AUTHOR]

The paper considers a number of key aspects surrounding the appropriation of the word gender into the Bulgarian language. "Gender theory", "gender ideology", "genderism", "genderization", "genderette" are only a few of the doublets which have become part of Bulgaria's political as well as everyday discourses. Our main goal is to analyze the recent developments by which these predominantly pejorative meanings and usages of the word have gained legitimacy, in part, with the aid of the official Bulgarian institutions. To do this, we focus on the definitions of "gender", which are offered in the recently published "Dictionary of new words in the Bulgarian language (from the first two decades of the 21st century)" under the authority of the Institute for Bulgarian Language at the Bulgarian Academy of Sciences. To set the context, we begin with an overview of the so-called anti-gender mobilizations. This is followed with an outline of the reception of the concept of "gender" in Bulgaria. Here we focus on two key fields. First, we look at the word's history within the Bulgarian academic space after the fall of the Iron Curtain, and second – the current political appropriations which fill the term with derogatory meanings. From there we continue with a critical examination of the explanatory-descriptive segments and the accompanying illustrative materials of the dictionary articles related to the word and its derivatives. Our main goal is to follow the shifts of meanings which the concept goes through. To further exemplify this, we draw a comparison with the previous version of the dictionary, which relies on a completely different explanatory strategy. [ABSTRACT FROM AUTHOR]

In this paper, we investigate the issue of hate speech by presenting a novel task of translating hate speech into non-hate speech text while preserving its meaning. As a case study, we use Spanish texts. We provide a dataset and several baselines as a starting point for further research in the task. We evaluated our baseline results using multiple metrics, including BLEU scores. We used a cross-validation approach and an average of the metrics per fold for evaluation. We achieved a 0.236 sentenceBLEU score on four folds. This study aims to contribute to developing more effective methods for reducing the spread of hate speech in online communities. [ABSTRACT FROM AUTHOR]

Twitter has experienced a tremendous surge in popularity over recent years, establishing itself as a prominent social media platform with a large user base. However, with this increased usage, there has been a concerning rise in the number of individuals resorting to derogatory language and expressing their opinions in a demeaning manner toward others. This surge in hate speech has drawn significant attention to the field of sentiment analysis, which aims to develop algorithms capable of detecting and analyzing emotions expressed in social networks using intuitive approaches. This paper focuses on addressing the complex task of detecting hate speech and aggressive behavior while performing target classification. We explored various deep-learning approaches, including LSTM, BiLSTM, CNN, and GRU. Each offers unique capabilities for capturing different aspects of the input data. We proposed an ensemble approach that combines the top three performing models. This ensemble approach benefits from the diverse strengths of each individual model showing F1 score of 0.85 for English-HS, 0.94 for English-TR, 0.92 for English-AB, 0.84 for Spanish-HS, 0.86 for Spanish-TR, 0.97 for Spanish-AB, 0.74 for multilingual-HS, 0.94 for multilingual-TR, and 0.88 for multilingual-AB. [ABSTRACT FROM AUTHOR]

Social media platforms have surpassed cultural and linguistic boundaries, thus enabling online communication worldwide. However, the expanded use of various languages has intensified the challenge of online detection of hate speech content. Despite the release of multiple Natural Language Processing (NLP) solutions implementing cutting-edge machine learning techniques, the scarcity of data, especially labeled data, remains a considerable obstacle, which further requires the use of semisupervised approaches along with Generative Artificial Intelligence (Generative AI) techniques. This paper introduces an innovative approach, a multilingual semisupervised model combining Generative Adversarial Networks (GANs) and Pretrained Language Models (PLMs), more precisely mBERT and XLM-RoBERTa. Our approach proves its effectiveness in the detection of hate speech and offensive language in Indo-European languages (in English, German, and Hindi) when employing only 20% annotated data from the HASOC2019 dataset, thereby presenting significantly high performances in each of multilingual, zero-shot crosslingual, and monolingual training scenarios. Our study provides a robust mBERT-based semisupervised GAN model (SS-GAN-mBERT) that outperformed the XLM-RoBERTa-based model (SS-GAN-XLM) and reached an average F1 score boost of 9.23% and an accuracy increase of 5.75% over the baseline semisupervised mBERT model. [ABSTRACT FROM AUTHOR]

Social media platforms have emerged as fertile ground for the proliferation of hate speech, which can exacerbate the dissemination of hate crimes. The Federal Bureau of Investigation UCR Program gathers data on hate crimes and disseminates annual reports to identify national patterns and inform law enforcement agencies and policymakers, these reports often fail to keep pace with urgent demands. Real-time monitoring and predictive analysis of hate crime trends are imperative for more effective prevention and response efforts. This paper presents a framework that leverages information extraction techniques to extract incidents from articles published in The New York Times, enabling accurate prediction of hate crime trends at both the federal and state levels. Experimental findings demonstrate the superiority of our approach compared to other traditional methods. By expanding forecasting approaches for federal and state levels' hate crime trends, this framework offers valuable insights for law enforcement agencies and policymakers. [ABSTRACT FROM AUTHOR]

The paper presents a novel data augmentation-based approach to develop explainable, deep learning models for hate speech detection. Hate speech is widely prevalent on online social media but difficult to detect automatically due to challenges of natural language processing and complexity of hate speech. Further, the decisions of the existing solutions possess constrained explainability since limited annotated data are available for training and testing of models. Therefore, this work proposes the use of text-based data augmentation for improving the performance and explainability of deep learning models. Techniques based on easy data augmentation, bidirectional encoder representations from transformers and back translation have been utilized for data augmentation. Convolutional neural networks and long short-term memory models are trained with augmented data and evaluated on two publicly available datasets for hate speech detection. Methods of LIME and integrated gradients are used to retrieve explanations of the deep learning models. A diagnostic study is conducted on test samples to check for improvement in the models as a result of the data augmentation. The experimental results verify that the proposed approach improves the explainability as well as the accuracy of hate speech detection. [ABSTRACT FROM AUTHOR]

With an aim to eliminate or reduce the spread of hate content across social media platforms, the development of artificial intelligence supported computational predictors is an active area of research. However, diversity of languages hinders development of generic predictors that can precisely identify hate content. Several language-specific hate speech detection predictors have been developed for most common languages including English, Chinese and German. Specifically, for Urdu language a few predictors have been developed and these predictors lack in predictive performance. The paper in hand presents a precise and explainable deep learning predictor which makes use of advanced language modelling strategies for the extraction of semantic and discriminative patterns. Extracted patterns are utilized to train an attention-based novel classifier that is competent in precisely identifying hate content. Over coarse-grained benchmark dataset, the proposed predictor significantly outperforms state-of-the-art predictor by 8.7% in terms of accuracy, precision and F1-score. Similarly, over fine-grained dataset, in comparison with state-of-the-art predictor, it achieves performance gain of 10.6%, 17.6%, 18.6% and 17.6% in terms of accuracy, precision, recall and F1-score. [ABSTRACT FROM AUTHOR]

This paper analyses online racial hate speeches in Portugal. The theory underpinning this research challenges the idea that racism is an epiphenomenon in European and Portuguese history. Beginning with a theoretical framework which conceives racism as a historically constructed system of power and privilege, we argue that its structural nature paved the way for the consolidation and normalisation of openly racist discourses. Moreover, this paper demonstrates that the nature of racist narratives can only be understood from a theoretical conceptualisation which problematises the Portuguese colonial past and its current continuities. In this research, a critical analysis of discourses produced and reproduced on social media, namely Facebook and Twitter, was carried out. From a comparative analysis of the periods before and during the COVID-19 pandemic, the idea that this health crisis coincided with a shift of the dominant racist narratives is thus challenged. [ABSTRACT FROM AUTHOR]

'Counterspeech' is often presented as a way in which individual citizens can respond to harmful speech while avoiding the potentially coercive and freedom-damaging effects of formal speech restrictions. But counterspeech itself can also undermine freedom by contributing to forms of social punishment that manipulate a speaker's choice set in uncontrolled ways. Specifically, and by adopting a republican perspective, this paper argues that certain kinds of counterspeech can dominate when they contribute to unchecked social norms that enable others to interfere arbitrarily with speakers. The presence of such domination can pose just as much a threat to freedom of speech as unchecked formal restrictions by threatening an individual's discursive status, revealing a problem for those who defend counterspeech as a freedom-protecting alternative. Rather than rejecting both counterspeech and legislation outright, however, this paper argues that the republican principle of parsimony ought to be exercised when deciding on appropriate harmful speech response. While the principle of parsimony allows for suitably-checked formal punishment for some of the most egregious forms of harmful speech, citizen-led counterspeech must be guided by a reliable set of norms against the use of social punishment where those who do engage in social punishment face certain costs. The presence of robust, widely-known, and reliable norms thus supports both formal and informal responses to harmful speech while maintaining a secure discursive status for all. [ABSTRACT FROM AUTHOR]

The paper's main goal is to compare laws banning hate speech with counterspeech as an effective method of curtailing hate speech. In the first part, the paper discussed three normative justifications for hate speech bans. Firstly, the line of argument developed by critical race theorists that assumes that hate speech leads to the direct harm and violation of individuals' rights. Secondly, paper examines the Weimar model that rests on the assumption that hate speech can lead to indirect harm to members of vulnerable minorities by creating a toxic environment, which opens the door to discrimination and even violence. Thirdly, the justification which is derived from the idea that such forms of extreme public speech violate the basic values and principles – such as inclusiveness, equality and mutual respect - on which constitutional democracies are built. This approach extends its argument from general values to the status of citizens by arguing that hate speech violates the equal standing of citizens by attempting to exclude certain members of society from the process of democratic deliberation based on their ascriptive characteristics. The second part of the paper looks at counterspeech as an efficient approach to limit harms that arise from hate speech. The author argues that legal bans should be reserved only for the most extreme forms of hateful rhetoric, while counterspeech is a valid option for tackling all other forms of hate speech. [ABSTRACT FROM AUTHOR]

Purpose: There is a scarcity of research studies on why people remain inactive when encountering and recognising misinformation online. The main aim of this paper is to provide a groundwork for future research into why users do not challenge misinformation on digital platforms by generating hypotheses through a synthesis of pertinent literature, including organisational behaviour, communication, human-computer interaction (HCI), psychology and education. Design/methodology/approach: Given the lack of directly related literature, this paper synthesised findings from relevant fields where the findings might be relevant, as the tendency to withhold opinions or feedback is a well-documented practice in offline interaction. Findings: Following the analysis of relevant literature, the potential reasons for online silence towards misinformation can be divided into six categories: self-oriented, relationship-oriented, others-oriented, content-oriented, individual characteristics and technical factors. Originality/value: Although corrections coming from peers can effectively combat misinformation, several studies showed that people in cyberspace do not take such action. To the best of the authors' knowledge, there has been scarce and virtually non-existent research investigating why people refrain from challenging others who post misinformation online. Thus, this paper attempts to address this gap and identify reasons in adjacent domains. The reasons provide a starting point for researching interventions to reduce reluctance and abstinence regarding the challenge of misinformation. The findings can be beneficial beyond the area of challenging misinformation and are extensible to other types of content and communication that people are hesitant to discuss and challenge, such as online injustice, prejudice and hate speech. [ABSTRACT FROM AUTHOR]

Machine-learning-based text classification is one of the leading research areas and has a wide range of applications, which include spam detection, hate speech identification, reviews, rating summarization, sentiment analysis, and topic modelling. Widely used machine-learning-based research differs in terms of the datasets, training methods, performance evaluation, and comparison methods used. In this paper, we surveyed 224 papers published between 2003 and 2022 that employed machine learning for text classification. The Preferred Reporting Items for Systematic Reviews (PRISMA) statement is used as the guidelines for the systematic review process. The comprehensive differences in the literature are analyzed in terms of six aspects: datasets, machine learning models, best accuracy, performance evaluation metrics, training and testing splitting methods, and comparisons among machine learning models. Furthermore, we highlight the limitations and research gaps in the literature. Although the research works included in the survey perform well in terms of text classification, improvement is required in many areas. We believe that this survey paper will be useful for researchers in the field of text classification. [ABSTRACT FROM AUTHOR]

Inspired by international experiences of the Sleeping Giants movement, since May 2020, Brazilian users of social networks such as Twitter have been denouncing advertisers and website funding platforms accused of disseminating disinformation and hate speech. This paper evaluates what criteria are presented to justify these demonetization campaigns, considering how those responsible for the Sleeping Giants Brazil accounts explain their choices of targets and what they consider to be fake news and offensive speech. The paper evaluates how the movement responds to criticism and questioning from social network users, clarifying their choices, publishing content that works as explanations of their methods. These justifications are presented as educational materials to expose to the public the difficulties in defining limits to free speech on digital networks. [ABSTRACT FROM AUTHOR]

The amount of outrage language, hate speech and strong language in digital communication is steadily growing, but academic studies on how to deal linguistically with anger and outrage are still largely lacking. Social science studies focus mainly on behavioral actions, while only a few investigate situations in which aggression is successfully handled verbally. On the other hand, complaint handling via webcare has received considerable scholarly attention, but its focus was never on de-escalation techniques. This paper explores what verbal de-escalation techniques can be learned from webcare employees who succeeded in satisfying complainants. More specifically, it combines insights from the (im)politeness framework and conversational human voice approach (CHV) to show that webcare practitioners have developed a tacit consensus on how to use language to deal with complaints. In a study with a mixed methods approach, a Twitter (now X) sample of responses to complaints is explored to identify linguistic and rhetorical features that encourage de-escalation. Thirteen features, that can be grouped into three categories, contributed to mollify the initial complainant. They can be characterized as markers of empathy (1), authenticity (2) and verbal accommodation (3) and they may help to calm heated debates in domains other than webcare. This article extends previous research by merging both pragma-linguistic and corporate communication taxonomies of complaint handling routines, adding forms of stylistic internal modification as possibly successful de-escalation features. • Webcare policy on complaint handling serves to uncover de-escalation techniques for outrage language. • Linguistic and rhetorical features that create empathy, authenticity and accommodation help to defuse heated Twitter discussions. • The uptake of responses to complaints helps to distinguish between successful and unsuccessful complaint handling. [ABSTRACT FROM AUTHOR]

Exposure to hate speech (HS) leads to desensitization of listeners. Yet, most evidence of this process has been obtained using self‐report measures. In this paper, we examined desensitization to HS using an unobtrusive, psychophysiological measure. In an experimental electrocardiogram study (N = 56), we observed heart rate (HR) deceleration after reading comments that contained HS. This suggested a substantive psychophysiological reaction of participants to hateful comments. However, such HR deceleration was not observed among participants preexposed to HS. People exposed to hateful comments thus appeared to show different HR responses to HS compared to people who were not previously exposed to such comments. Consequently, not only does frequent exposure to HS influence an individual's beliefs as observed in earlier studies, but it also impacts psychophysiological reactions to derogatory language. [ABSTRACT FROM AUTHOR]

Supervised training has traditionally been the cornerstone of hate speech detection models, but it often falls short when faced with unseen scenarios. Zero and few-shot learning offers an interesting alternative to traditional supervised approaches. In this paper, we explore the advantages of zero and few-shot learning over supervised training, with a particular focus on hate speech detection datasets covering different domains and levels of complexity. We evaluate the generalization capabilities of generative models such as T5, BLOOM, and Llama-2. These models have shown promise in text generation and have demonstrated the ability to learn from limited labeled data. Moreover, by evaluating their performance on both Spanish and English datasets, we gain insight into their cross-lingual applicability and versatility, thus contributing to a broader understanding of generative models in natural language processing. Our results highlight the potential of generative models to bridge the gap between data scarcity and model performance across languages and domains. [ABSTRACT FROM AUTHOR]

Offensive content, such as verbal attacks, demeaning comments, or hate speech, has become widespread on social media. Automatic detection of this content is considered an important and challenging task. Although several research works have been proposed to address this challenge for high-resource languages, research on detecting offensive content in Dialectal Arabic (DA) remains under-explored. Recently, the detection of offensive language in DA has gained increasing interest among researchers in Natural Language Processing (NLP). However, only a limited number of annotated datasets have been introduced for single or multiple coarse-grained dialects. In this paper, we introduce Offensive Moroccan Comments Dataset (OMCD), the first dataset for offensive language detection for the Moroccan dialect. First, we present the data collection steps, the statistical analysis, and the annotation guidelines of the introduced dataset. Then, we evaluate several state-of-the-art Machine Learning (ML) and Deep Learning (DL) based models on the OMCD dataset. Finally, we highlight the impact of emojis on the evaluated models for offensive language detection. [ABSTRACT FROM AUTHOR]

The goal of hate speech detection is to filter negative online content aiming at certain groups of people. Due to the easy accessibility and multilinguality of social media platforms, it is crucial to protect everyone which requires building hate speech detection systems for a wide range of languages. However, the available labeled hate speech datasets are limited, making it difficult to build systems for many languages. In this paper we focus on cross-lingual transfer learning to support hate speech detection in low-resource languages, while highlighting label issues across application scenarios, such as inconsistent label sets of corpora or differing hate speech definitions, which hinder the application of such methods. We leverage cross-lingual word embeddings to train our neural network systems on the source language and apply them to the target language, which lacks labeled examples, and show that good performance can be achieved. We then incorporate unlabeled target language data for further model improvements by bootstrapping labels using an ensemble of different model architectures. Furthermore, we investigate the issue of label imbalance in hate speech datasets, since the high ratio of non-hate examples compared to hate examples often leads to low model performance. We test simple data undersampling and oversampling techniques and show their effectiveness. [ABSTRACT FROM AUTHOR]

Many activities are conducted on social media platforms, such as promoting products, sharing news and sharing achievements. As a result of users' freedom and anonymity on social media platforms, hate speech and harassment are common. Posts that spread hate and offense should be detected and deleted as soon as possible as they spread very quickly and have many negative consequences for the human race. The detection task becomes significantly more difficult when online users use code-mixed text in places where English is not the primary language. The problem of hate speech detection has recently been reduced to a binary classification task, without taking into account its topical focus and its target-oriented nature. Because there is no combined annotated dataset and scientific study that can provide insight into the relationship between offense traits, existing techniques usually only examine one or two offense traits at a time. Furthermore, these techniques are not efficient for multilingual, where most conversations are code-mixed. In this paper, we propose an optimal feature extraction and hybrid diagonal gated recurrent neural network (FE-DGRNN) for hate speech detection and sentiment analysis in multilingual code-mixed texts. The proposed FE-DGRNN technique consists threefold processes. After preprocessing, we first introduce an improved seagull optimization (ISO) algorithm for multiple features extraction from given code-mixed texts. Then, we utilize a quantum search optimization algorithm to optimize the extracted features which reduces the data dimensionality issues in further detection phase. Next, a hybrid diagonal gated recurrent neural network (Hyb-DGRNN) introduces to detect hate speech and analyzes sentiment on their language. In order to validate the effectiveness of our proposed FE-DGRNN technique, we conducted experiments on the HASOC 2019 dataset. This dataset includes posts written in English, Hindi and German, allowing us to evaluate the performance of our approach across multiple languages. From the simulation results, we observed that the accuracy of FE-DGRNN is 87.74%, 88.98% and 84.74% for Task-1, Task-2 and Task-3, respectively, for multilingual code-mixed texts dataset. Overall, the proposed FE-DGRNN technique shows a significant improvement in accuracy, precision, recall and F-measure compared to other classifiers, indicating its potential to be a robust and effective tool for hate speech detection and sentimental analysis in multilingual code-mixed texts. [ABSTRACT FROM AUTHOR]

This paper analyzes the discourse concerning Islam and Muslims by assessing the extreme right-wing party, Vox, on Twitter. In addition, this paper examines the incidence (impact and reactions) of this party on the users of this social network. The objectives of this study are as follows: to identify themes and topics concerning this discourse; to analyze how the discourse is articulated and represented; and to understand the impact of this discourse by measuring the engagement of the most viral publications. To do so, we observed the publications posted by the party via its official account throughout 2022. The research methodology was based on qualitative and quantitative content analysis, and the publications themselves were monitored to ascertain the level of engagement. The results of this study clearly show that Vox created Islamophobic narratives; thus, Islam and Muslims are explicit targets of Vox's hate speech. Their rhetoric referred to security threats and threats to national identity, with the recurrent use of the idea that there is a "danger of Islamization" in Spain. The party uses disinformation and hoaxes, and users respond in a polarized way. The results of this study alert us to the worrying levels of radicalization and the normalization of Islamophobic racist discourse in the examined context. [ABSTRACT FROM AUTHOR]

Hate speech is a contextual phenomenon. What offends or inflames in one context may differ from what incites violence in a different time, place, and cultural landscape. Theories of hate speech, especially Susan Benesch's concept of "dangerous speech" (hateful speech that incites violence), have focused on the factors that cut across these paradigms. However, the existing scholarship is narrowly focused on situations of mass violence or societal unrest in America or Europe. This paper discusses how online hate speech may operate differently in a postcolonial context.¹ While hate speech impacts all societies, the global South--Africa in particular--has been sorely understudied. I posit that in postcolonial circumstances, the interaction of multiple cultural contexts and social meanings form concurrent layers of interpretation that are often inaccessible to outsiders. This study expands the concept of online harms by examining the political, social, and cultural dimensions of data-intensive technologies. The paper's theories are informed by fieldwork that local partners and I conducted in Kasese, Uganda in 2019-2020, focusing on social unrest and lethal violence in the region following the 2016 elections. The research, completed with assistance from the Berkeley Human Rights Clinic, included examining the background and circumstances of the conflict; investigating social media's role in the conflict; designing a curriculum around hate speech and disinformation for Ugandan audiences; creating a community-sourced lexicon of hateful terms; and incorporating community-based feedback on proposed strategies for mitigating hate speech and disinformation. I begin this with a literature review of legal theory around hate speech, with a particular focus on Africa, and then turn to the legal context around hate speech and social media use in Uganda, examining how the social media landscape fueled past conflicts. Then I explain my Kasese fieldwork and the study's methodology, before describing initial results. I follow with a discussion of applications to industry, specifically how hate speech is defined and treated by Meta's Facebook, the dominant social media provider in Kasese. It progresses to a discussion of the implications of the study results and legal and policy recommendations for technology companies stemming from these findings. Importantly, I apply the research findings to expand existing scholarship by proposing a new sixth "hallmark of dangerous speech" to augment Benesch's paradigm. Adding "calls for geographic exclusion" as a new qualifier for dangerous speech stems from the particular characteristics embodied by postcolonial hate speech. Examples from the Kasese study illustrate how this phenomenon upends platforms' expectations of hate speech--which may not consider "Coca-Cola bottle" to be an epithet. The application of this new hallmark will create a more inclusive understanding of hate speech in localized contexts. This paper's conclusions and questions may challenge platforms that must address hate speech and content moderation at a global scope and scale. It will examine the prevalence and role of social media platforms in Africa, and how these platforms have provided resources and engagement with civil society in these regions. [ABSTRACT FROM AUTHOR]

Little is known about the current state of research on the involvement of young people in hate speech. Thus, this systematic review presents findings on a) the prevalence of hate speech among children and adolescents and on hate speech definitions that guide prevalence assessments for this population; and b) the theoretical and empirical overlap of hate speech with related concepts. This review was guided by the Cochrane approach. To be included, publications were required to deal with real-life experiences of hate speech, to provide empirical data on prevalence for samples aged 5 to 21 years and they had to be published in academic formats. Included publications were full-text coded using two raters (κ =.80) and their quality was assessed. The string-guided electronic search (ERIC, SocInfo, Psycinfo, Psyndex) yielded 1,850 publications. Eighteen publications based on 10 studies met the inclusion criteria and their findings were systematized. Twelve publications were of medium quality due to minor deficiencies in their theoretical or methodological foundations. All studies used samples of adolescents and none of younger children. Nine out of 10 studies applied quantitative methodologies. Eighteen publications based on 10 studies were included. Results showed that frequencies for hate speech exposure were higher than those related to victimization and perpetration. Definitions of hate speech and assessment instruments were heterogeneous. Empirical evidence for an often theorized overlap between hate speech and bullying was found. The paper concludes by presenting a definition of hate speech, including implications for practice, policy, and research. [ABSTRACT FROM AUTHOR]

Faced with hate speech, oppressed groups can use their own speech to respond to their verbal oppressors. This "counterspeech," however, sometimes itself takes on a hateful form. This paper explores the moral standing of such "hateful counterspeech." Is there a fundamental moral asymmetry between hateful counterspeech, and the hateful utterances of dominant or oppressive groups? Or are claims that such an asymmetry exists indefensible? I argue for an intermediate position. There is a key moral asymmetry between these two forms of speech. But, this asymmetry notwithstanding, hateful counterspeech is capable of enacting serious harms—and so, contrary to what many legal theorists have argued, it is in principle an appropriate object of legal regulation. I begin by considering the central argument for thinking that hateful counterspeech is not seriously troubling. This argument holds that oppressed groups lack authority—and, by extension, "speaker power." Yet this argument, I suggest, sits in tension with the fact that low-status members of dominant groups can, through their utterances, seriously harm members of oppressed groups. Philosophers of language have developed sophisticated arguments to explain this last phenomenon: they have argued that speaker power is relativised to particular jurisdictions; that it can be acquired dynamically in local settings; and that it is socially dispersed. I argue that, in light of these arguments, it appears that hateful counterspeech, too, can generate serious harms. Nevertheless, I show that this conclusion is compatible with recognising a crucial moral asymmetry between hateful counterspeech and the hate speech of oppressors. [ABSTRACT FROM AUTHOR]

A growing body of work in free speech theory is interested in the nature of counterspeech, i.e. speech that aims to counteract the effects of harmful speech. Counterspeech is usually defined in opposition to legal responses to harmful speech, which try to prevent such speech from occurring in the first place. In this paper we challenge this way of carving up the conceptual terrain. Instead, we argue that our main classificatory division, in theorising responses to harmful speech, should be between pro-discursive and anti-discursive responses. Some legal responses to harmful speech, so we argue, make a positive discursive contribution in their own right. That is, legal restrictions on harmful speech can have a function that is importantly similar to speech that aims at countering the effects of harmful speech. [ABSTRACT FROM AUTHOR]

In this paper, we identify and examine an overlooked strategy to counter bigoted speech on the spot. Such a strategy we call 'bending'. To 'bend', in our sense, is to deliberately give a distorted response to a speaker's harmful move – precisely, an ameliorative response, which may turn that move into a different, less harmful, contribution. To substantiate our proposal, we distinguish two ideas of uptake – interpretation and response – and argue for the general claim that a distorted response on the hearer's part may end up transforming a speaker's contribution. Patterns of distortion have been analyzed in the literature as unjustly undermining speakers' agency and exacerbating oppression. Our analysis shows that, under certain circumstances, distortion can be employed to derail bigoted speech and thus serve the purposes of social justice. We close by discussing the virtues and limits of bending vis-à-vis a different, much-discussed, counterspeech strategy, i.e. 'blocking' (Langton 2018). [ABSTRACT FROM AUTHOR]

The evolution of social media and online platforms created more opportunities to express one's thoughts without any burden. This emergence paved the way to construct more intelligent systems but by taking advantage of these platforms some people display their vengeance by harassing or abusing comments. The attackers leak sensitive information about an individual or an organization through the internet using abusive words, hate comments, threatening information, fake news, etc. So, online shaming detection with an optimum classification mechanism becomes necessary to avoid such negative consequences. In this paper, a novel online shaming detection mechanism using a Convolutional Neural Network Bidirectional Gated Recurrent Unit-based Hybrid Archimedes Tasmanian Devil (CNNBiGRU-HATD) approach is proposed to detect and classify the offensive comments and non-offensive comments. If the comment seems offensive, it is further classified into six categories according to shaming types such as abusive, doxing, Passing Judgment, Ethnic/Religious, Joke/Sarcasm, and Whataboutery. The utilized four Twitter-based datasets are initially preprocessed and feature extracted to improve the classification performance of the CNNBiGRU-HATD approach. After accurate classification, the information from the classified shaming types is collected and authenticated. The proposed approach is simulated using the Python platform. To validate the potential capability of the proposed CNNBiGRU-HATD approach, it is related to the existing methods by various parameters. The experimental result reveals that the proposed CNNBiGRU-HATD method achieves higher accuracy of 96.5% for the labeled hate speech detection dataset. [ABSTRACT FROM AUTHOR]

Hate Speech Detection (HSD) aims to identify whether a text contains hate speech content, which often refers to discrimination and is even associated with a hate crime. The mainstream methods jointly train the HSD problem with relevant auxiliary problems, e.g., emotion detection and sentiment analysis, under the paradigm of Multi-Task Learning (MTL). In this paper, we improve HSD by integrating it with emotion detection, since we take inspiration from the potential correlations between hate speech and certain negative emotion states, which have been studied theoretically and empirically. To be specific, we can concatenate their hateful labels and predicted emotion states as pseudo-multiple labels for hate speech samples, formulating a pseudo-Multi-Label Learning (MLL) problem. Beyond the existing MTL-HSD methods, we further incorporate this pseudo-MLL problem and solve it by capturing the correlations between hate speech and negative emotion states, so as to improve the performance of HSD. Based on these ideas, we propose a novel HSD method named the E motion-correlated H ate S peech Detect OR (EHS or). We conduct extensive experiments to evaluate EHS or , and the results show that it can consistently outperform the existing HSD methods across benchmark datasets. • Validating correlations between hate speech and specific negative emotions. • Proposing a new MTL-based HSD method EHSOR from the perspective of MLL. • EHSOR empirically outperforms baseline methods over three HSD benchmarks. [ABSTRACT FROM AUTHOR]

The aim of this paper is to analyse hate speech in 19th century Germany from a linguistic perspective. The paper thus examines the works of three German authors who are regarded as exemplary for the emergence and further development of German-language hate speech: 1) Ernst Moritz Arndt, Über den Volkshaß und über den Gebrauch einer fremden Sprache (1813); 2) Wilhelm Marr (1819-1904), Der Sieg des Judenthums über das Germanenthum (1879), and 3) Otto Böckel (1859-1923), Die Juden - Die Könige unserer Zeit (1887). The analysis shows that the hate language had been particularly present in journalism and literature since the end of the 18th century, affecting diverse aspects of social life and becoming part of the political debate, especially during the emergence of the German nation after the Napoleonic wars. [ABSTRACT FROM AUTHOR]

In the 21st century – the time of the Internet, hate speech is present throughout social media. Hate speech existed even before the advent of the Internet. However, what is different about online hate speech is the speed at which it is transmitted on certain social media, such as Twitter and Facebook. In this sense, the question of whether and to what extent online hate speech should, or rather should not be protected was posed. This paper aims to explore whether online hate speech deserves freedom of expression protection. Public officials, civil servants, and politicians often use different social media platforms, especially Twitter, to communicate new political ideas, upcoming events, and even conspiracy theories. In this sense, the aim of the paper is to investigate whether a need to balance between constitutionally protected freedom of expression and the core values of democratic society arose from the development of Internet technology. [ABSTRACT FROM AUTHOR]

Information literacy emerged in the 1970s to address information needs in an active and critical way. However, the increasing availability of information and the evolution of technology led information literacy to focus more on instrumental aspects and less on its cognitive, axiological, and critical dimensions. As a response, a new wave emerged to explicitly focus on the most critical and emancipatory aspects: Critical Information Literacy (CIL). This concept has been developed in a dynamic and spontaneous way through the various practices and publications on the topic and as a reflection of the social advances in recent decades. In this paper, CIL is presented as a means of action for informational activism to promote changes in society aligned with the values of human rights and social justice. We present a review of the main authors that have worked with CIL, including the active scene in Brazil and its Freirean influence, aiming at systematizing the concept and its main characteristics in a global landscape dominated by misinformation and absence of informational and digital competences. The concept of informational activism is presented and discussed as one of the main lines of action for Library and Information Science in today's polarized society. [ABSTRACT FROM AUTHOR]

Through addressing four public universities in Egypt, the author of this paper aims to uncover the main barriers women faculty face when fulfilling their academic roles. The author employed a qualitative research method through conducting semi‐structured interviews with 40 female academics. Moreover, he subsequently used thematic analysis to determine the main ideas in the transcripts. The author of the present study has identified the poor representation of women faculty at both professorial levels and in managerial academic positions. The study indicated that heavy teaching loads, lengthy working hours, accusations of low productivity, growth of masculine power distance and limited empowerment opportunities are the main barriers hindering the status and representation of women faculty in the addressed universities. The findings also showed that social realities, which include the implicit relations among humans (male and female faculty in this case) in the workplace and also relations between humans and their surrounding natural world (stakeholders in the surrounding environment in the present case) decide the volume of knowledge and level of power women might attain. This paper contributes by filling a gap in HR management and higher education, in which empirical studies on the status of women faculty and the barriers they face have been limited so far. [ABSTRACT FROM AUTHOR]

Social media is a more common and powerful platform for communication to share views about any topic or article, which consequently leads to unstructured toxic, and hateful conversations. Curbing hate speeches has emerged as a critical challenge globally. In this regard, Social media platforms are using modern statistical tools of AI technologies to process and eliminate toxic data to minimize hate crimes globally. Demanding the dire need, machine and deep learning-based techniques are getting more attention in analyzing these kinds of data. This survey presents a comprehensive analysis of hate speech definitions along with the motivation for detection and standard textual analysis methods that play a crucial role in identifying hate speech. State-of-the-art hate speech identification methods are also discussed, highlighting handcrafted feature-based and deep learning-based algorithms by considering multimodal and multilingual inputs and stating the pros and cons of each. Survey also presents popular benchmark datasets of hate speech/offensive language detection specifying their challenges, the methods for achieving top classification scores, and dataset characteristics such as the number of samples, modalities, language(s), number of classes, etc. Additionally, performance metrics are described, and classification scores of popular hate speech methods are mentioned. The conclusion and future research directions are presented at the end of the survey. Compared with earlier surveys, this paper gives a better presentation of multimodal and multilingual hate speech detection through well-organized comparisons, challenges, and the latest evaluation techniques, along with their best performances. [ABSTRACT FROM AUTHOR]

We introduce a Fuzzy Property Grammar System (FPGS), a formalism that integrates a Fuzzy Property Grammar into a linguistic grammar system to formally characterize metaphorical evaluative expressions. The main scope of this paper is to present the formalism of FPGS and to show how it might provide a formal characterization of hate speech linguistic evaluative expressions with metaphors (as fuzzy concepts), together with evaluating their degree of linguistic violence. Linguistic metaphors are full of semantic coercions. It is necessary to formally characterize the context of the communication to acknowledge the extralinguistic constraints of the pragmatic domain, which establishes whether an utterance is violent. To show the applicability of our formalism, we present a proof of concept. By compiling and tagging a 3000-tweet corpus, we have extracted a lexicon of hate speech metaphors. Furthermore, we show how FPGS architecture can deal with different types of hate speech and can identify implicit violent figurative evaluative expressions by context and type. Although we are still in the experimental phase of our project and cannot present conclusive results at the computational level, the proof-of-concept results show that our formalism can achieve the desired outcome. [ABSTRACT FROM AUTHOR]

This paper argues that the binary opposition in the treatment of hate speech in the US and Europe hides non-binary preoccupations that reflect different primary fears which do not fall along the same 'scale'. European liberal democracies fear the consequences of hate speech being left uncensored in the public domain (a WHAT concern) whilst America fears the consequences of content interventions by government (a WHO concern). The paper then proposes that the German Network Enforcement Law of 2017 builds a bridge between American and European speech traditions. NetzDG requires major platforms to moderate content in response to user takedown notices based on legally imposed speech standards. The mechanism of public standards being enforced through private processes is arguably uniquely adept at simultaneously assuaging the primary European fear about the absence of effective speech controls in the public domain and the primary American fear about the presence of governmental censorship. [ABSTRACT FROM AUTHOR]

Fundamental values, such as human dignity, freedom, democracy, equality, and respect for human rights are essential values of a democratic, social and participatory society. Despite this, in the past few years, it is possible to see an increase in hate speech, racism and hate crimes in many countries of the EU. This paper aims to present the state-of-art of the Portuguese situation regarding to discrimination, hate speech and racism, based on literature review of national and international resources, and interviews with persons that have suffered from hate speech, as well as experts in the field. Results forecast a long pathway still to mitigate these phenomena and claim that it is fundamental to have an intervention that includes all societal actors: government, organisations, civil society and citizens. The researchers hope that this paper enriches the knowledge around the concepts discussed and that is useful for the development of strategies towards the promotion of a more equitable society. [ABSTRACT FROM AUTHOR]

Online toxic discourses could result in conflicts between groups or harm to online communities. Hate speech is complex and multifaceted harmful or offensive content targeting individuals or groups. Existing literature reviews have generally focused on a particular category of hate speech, and to the best of our knowledge, no review has been dedicated to hate speech datasets. This paper systematically reviews textual hate speech detection systems and highlights their primary datasets, textual features, and machine learning models. The results of this literature review are integrated with content analysis, resulting in several themes for 138 relevant papers. This study shows several approaches that do not provide consistent results in various hate speech categories. The most dominant sets of methods combine more than one deep learning model. Moreover, the analysis of several hate speech datasets shows that many datasets are small in size and are not reliable for various tasks of hate speech detection. Therefore, this study provides the research community with insights and empirical evidence on the intrinsic properties of hate speech and helps communities identify topics for future work. [ABSTRACT FROM AUTHOR]

Hate Speech is a frequent problem occurring among Internet users. Recent regulations are being discussed by U.K. representatives ("Online Safety Bill") and by the European Commission, which plans on introducing Hate Speech as an "EU crime". The recent legislation having passed in order to combat this kind of speech places the burden of identification on the hosting websites and often within a tight time frame (24 h in France and Germany). These constraints make automatic Hate Speech detection a very important topic for major social media platforms. However, recent literature on Hate Speech detection lacks a benchmarking system that can evaluate how different approaches compare against each other regarding the prediction made concerning different types of text (short snippets such as those present on Twitter, as well as lengthier fragments). This paper intended to deal with this issue and to take a step forward towards the standardization of testing for this type of natural language processing (NLP) application. Furthermore, this paper explored different transformer and LSTM-based models in order to evaluate the performance of multi-task and transfer learning models used for Hate Speech detection. Some of the results obtained in this paper surpassed the existing ones. The paper concluded that transformer-based models have the best performance on all studied Datasets. [ABSTRACT FROM AUTHOR]

Purpose: The purpose of this article is to draw attention to how harmful and inaccurate discourses pertaining to disaster responsibility is produced, the negative implications such narratives pose and the role of the media in the ways in which discourses about queerness and disaster are reported. Design/methodology/approach: Throughout this paper, the authors detail examples of media reporting on discourses relating to people with diverse sexual orientation, gender identity, gender expression and sex characteristics (SOGIESC) being blamed and held responsible for disasters across the world. The authors examine the value of such reporting as well as describing the harm blame narratives have on queer people and communities. Findings: There is little value in reporting on accounts of people publicly declaring that people with diverse SOGIESC are to blame for disaster. More sensitivity is needed around publishing on blame discourses pertaining to already marginalised communities. Originality/value: This article contributes to the developing scholarship on lesbian, gay, bisexual, transgender, queer, intersex, agender, asexual and aromantic individuals, plus other gender identities and sexual orientations (LGBTQIA+/SOGIESC) and disasters by detailing the harm of blame discourses as well as drawing attention to how the media have a role to play in averting from unintentionally providing a platform for hate speech and ultimately enhancing prejudice against people with diverse SOGIESC. [ABSTRACT FROM AUTHOR]

The anonymity of the Manosphere has provided a fertile breeding ground for the spread of misogyny through increased hate speech on the Internet. In recent years, this virtual space composed mostly of men, has been the subject of numerous studies aimed at identifying the discursive mechanisms of identity constructions that underlie the denigration of women. This paper takes the findings from international research on the subject to address a phenomenon still little explored in Spain through the analysis of misogynistic comments on Burbuja.info. Economic forum. The analysis sample comprises 4,281 messages that include the lexeme “woman” posted over 2.5 years, 761 of which contain expressions that ridicule, disparage, or insult women. Thematic analysis demonstrates the repetition of misogynistic topics and tropes common in the most radical Reddit and 4chan subforums, such as hypergamy, objectification, and the constant disparagement of the world of women, not to mention criticism of feminism, left-wing political parties, and legislation on gender violence. The results obtained show that the “male identity crisis” is narratively constructed in the messages on Burbuja.info using the motif of sacrifice –the cornerstone of a monotypic story in which the purpose of the hero-man-victim’s journey is to punish the villain-woman. Furthermore, we also found that there are different configurations of identity related to attitudes toward women; these match up with the four most prominent masculinist subcultures identified in the leading literature on the subject: Men Rights activists (MRA), Men Going Their Own Way (MGTOW), Pickup Artists (PUA), and Involuntary Celibates (Incels). [ABSTRACT FROM AUTHOR]