Hate Speech Research Articles

AI Based Identification of Inappropriate Language

Abstract: In the current digital era, the widespread use of online communication has raised the demand for automated systems that can recognize and block hate speech, improper language, and objectionable information. The usefulness of three different machine learning algorithms—Decision Trees, Random Forest, Long Short-Term Memory (LSTM), and others—in determining if a given text contains objectionable language is investigated in this work. While LSTM and represent state-of-the-art deep learning algorithms capable of processing unstructured text input, Random Forest and Decision Trees are standard machine learning techniques that rely on organized feature engineering. This study compares different algorithms in an effort to shed light on their advantages and disadvantages in dealing with the changing demands of online content moderation. Our results show each model's accuracy, precision, recall, and F1-score.

Social sensing with big data: Detecting hate speech in social media

The internet's accessibility and social media platforms, like Facebook and Twitter, have accelerated the spread of hate speech and fake news, both of which can be detrimental to society's overall well-being. Identifying and tracking hate speech is becoming increasingly difficult for the public, private citizens, legislators, and academics. Despite efforts to leverage automatic detection and monitoring techniques, their performances are still far from satisfactory. This study employs Natural Language Processing (NLP) and Machine Learning (ML) approaches to detect hate speech for decision-making. The result showed that the Support Vector Machine (SVM) algorithm has the best performance with an accuracy of 0.86 compared to the Random Forest with 0.8 accuracy. The manual evaluation of the performance of our algorithm yielded an inter-annotator agreement Cronbach’s alpha (α = .775).

From hate speech to HateLess. The effectiveness of a prevention program on adolescents’ online hate speech involvement

Hate speech, a societal scourge, has found a potent conduit in modern technology, exacerbating its reach and impact. Adolescents are particularly vulnerable to the pernicious effects of online hate speech, which threaten their well-being and social cohesion. Addressing this critical issue, the current study uses a quasi-experimental design to evaluate the efficacy of “HateLess. Together Against Hatred” in reducing adolescents' online hate speech perpetration and victimization and increasing their online hate speech countering. A sample of 820 adolescents aged 12 to 16 (M = 13.27, SD = 1.04) from 11 German schools was divided into an intervention group (n = 567), who underwent the one-week anti-hate speech program, and a control group (n = 253). Utilizing multilevel mediation analysis, the study revealed that program participation significantly reduced online hate speech perpetration and victimization while enhancing countering hate speech. Notably, the intervention's success was partially mediated by increased empathy, which, in turn, predicted a decrease in hate speech perpetration and an increase in counter-speech, and by increased self-efficacy, which, in turn, was linked to reduced victimization and heightened countering hate speech. These promising outcomes suggest that HateLess could be a viable strategy for mitigating hate speech engagement and fostering a more respectful online discourse among adolescents.

Memes as symbols of hate speech

Los memes de Internet se han convertido en una expresión cultural y artística de las redes sociales más populares. Los memes son expresiones gráficas acompañadas de un pequeño texto que reflejan una idea o noticia de actualidad con cierto tono humorístico que se viralizan con gran facilidad. Este artículo pretende abordar el problema de la difusión del discurso del odio a través de los memes en las redes sociales. Esta difusión del discurso del odio a través de memes es especialmente relevante en política y se analizarán algunos ejemplos significativos de memes políticos para destacar su gran influencia en la celebración de elecciones.

The implications of the Arabic case for Bangla content moderation on Facebook: future considerations for combating ‘Bangla’ hate speech

The implications of the Arabic case for Bangla content moderation on Facebook: future considerations for combating ‘Bangla’ hate speech

Disinformation Dynamics Unveiling the Impact of Echo Chambers in Shaping Online Public Opinion

The proliferation of misinformation and the emergence of echo chambers in the online environment pose significant challenges to modern democracies, directly impacting public opinion and social behaviors. This study focuses on the analysis of a Facebook group centered around a prominent Romanian political figure, boasting 93,800 members and averaging ten daily posts. Using advanced machine learning and AI-based hate speech detection, the study uncovers systematic echo chamber construction and the amplification of misinformation. The findings emphasize the influence of online echo chambers on public opinion and underscore the need to maintain information integrity in the media landscape and communication. This research has important implications for scholars, policymakers, and media practitioners, indicating the critical need to address the challenges posed by misinformation and echo chambers in the online environment

Combating Hate Speech on Social Media: Applying Targeted Regulation, Developing Civil-Communicative Skills and Utilising Local Evidence-Based Anti-Hate Speech Interventions

Social media platforms such as Facebook and X (formerly Twitter) set their core aim as bringing people and communities closer together. Yet, they resemble a digital communicative battleground in which hate speech is increasingly present. Hate speech is not benign. It is the communicative driver of group oppression. It is therefore imperative to disarm this digital communicative battlefield by (a) regulating and redesigning social media platforms to prevent them from playing an active and enabling role in the dissemination of hate speech and (b) empowering citizen-users and local civil associations to recognise and actively counter hate speech. This top-down and bottom-up approach necessarily enforces responsibility and builds capacity. This requires that we adapt and combine three aspects of communicative peacebuilding: first, the (re)building of civil-communicative institutions; second, the use of digital citizenship educational programmes to support the development of civil-communicative skills for using social media; and third, the identification and use of local civil capacity and knowledge, which manifests in the present context in the use of local evidence-based anti-hate speech interventions. We argue that this interdisciplinary combinatorial approach has the potential to be effective because it combines two things: it places responsibility on relevant actors to both make social media safer and to navigate it harmlessly and responsibly; and it simultaneously helps build the capacity for actively identifying and countering hate speech in civil societies.

The impact of politicians’ behaviors on hate speech spread: hate speech adoption threshold on twitter in Japan

AbstractIn this paper, we hypothesized that a leading politician’s behavior signaling support for the ideas of hate groups will increase hate speech by shifting social norms. To examine the hypotheses, in particular, to quantify the social norms shift, the study utilized an adoption threshold measure based on Twitter retweet networks. Our empirical study focused on the hate speech spread effect of an announcement by Yuriko Koike, Governor of Tokyo, declining to participate in a memorial ceremony for Korean massacre victims. The results support the hypothesis: after Koike’s announcement, adoption thresholds of hate speech users were significantly lowered, while the number of hate speech users increased. Further, this study compared the effect of Koike’s announcement to the effect of a North Korean missile launch, a national security threat to Japan. The average hate speech adoption threshold was lower after Koike’s announcement than after the missile launch, suggesting that a leading politician’s behavior could have a greater impact on shifting norms of prejudice than even a nationally threatening event.

Semantic Conversational Content Moderation

In today's digital environment, social media platforms have become an integral part of our daily lives, facilitating global communication, knowledge sharing and community building. However, these platforms are increasingly vulnerable to the spread of offensive and toxic content, including misinformation, harassment and hate speech. Such content poses a significant threat to the safety and well-being of Internet users. In response to this immediate problem, we set out to develop an AI-based conversation moderation service aimed at effectively detecting and removing offensive or semantically toxic information in real time. Our solutions strive to make the internet safer and more user-friendly, thereby promoting a more positive and inclusive online environment. Key Words: Natural Language Processing(NLP), Moderation, Semantic Analysis, Mistral.

Instagram Fake Account Detection Based on Machine Learning

The popularity of social media continues to grow, and its dominance of the entire world has become one of the aspects of modern life that cannot be ignored. The rapid growth of social media has resulted in the emergence of ecosystem problems. Hate speech, fraud, fake news, and a slew of other issues are becoming un-stoppable. With over 1.7 billion fake accounts on social media, the losses have al-ready been significant, and removing these accounts will take a long time. Due to the growing number of Instagram users, the need for identifying fake accounts on social media, specifically in Instagram, is increasing. Because this process takes a long time if done manually by humans, we can now use machine learning to identify fake accounts thanks to the rapid development of machine learning. We can detect fake accounts on Instagram using machine learning by implementing the combination of image detection and natural language processing.

Party Politics and National Integration in Nigeria: An Appraisal of All Progressive Congress 2019-2023

This study examined party politics and national integration in Nigeria: an appraisal of All Progressive Congress (APC). It conceptualized political parties and national integration. The study raised two objectives namely: Ascertain how All Progressive Congress (APC) politics has promoted national integration in Nigeria between 2015 and 2023. Determine how All Progressive Congress (APC) politics has undermined national integration in Nigeria. Systems theory was used as the theory for the study. Ex-post facto design was used. Data used for the study were derived through secondary sources. Thematic content analysis was used to analyze the data. The findings of the study show that APC has made some contributions towards national integration in Nigeria. The finding of the study also shows that these efforts have their challenges which militated against them in the realization of their objectives. While the efforts were intended to achieve national integration, solidarity and unity, it ended up creating disunity among the people, and marginalization of one section by the other thereby leading to intense struggle for power between and or among people from different ethnic groups. The study concluded that due to the nature of party politics, APC as a political party has not been able to live up to expectation as a veritable instrument for national integration. Secondly, some of the party leaders are more powerful than the political party itself and as such their selfish personal interest overrides the interest of the party and the nation as the case may be. The study recommended that political parties must run issue-based campaigns devoid of hate speech or disinformation to restore their integrity in the eyes of citizens. Once elected and/ or appointed to political office, party members must follow through on the promises made during the campaign period. Also, political parties must be held accountable for their abuses of the system and failure to represent the interests of the Nigerian people. Keywords: Political Party, Party Politics, National Integration

The Dark Side of Metaverse: A Multi-Perspective of Deviant Behaviors From PLS-SEM and fsQCA Findings

The metaverse has created a huge buzz of interest because such a phenomenon is emerging. The behavioral aspect of the metaverse includes user engagement and deviant behaviors in the metaverse. Such technology has brought various dangers to individuals and society. There are growing cases reported of sexual abuse, racism, harassment, hate speech, and bullying because of online disinhibition make us feel more relaxed. This study responded to the literature call by investigating the effect of technical and social features through mediating roles of security and privacy on deviant behaviors in the metaverse. The data collected from virtual network users reached 1121 respondents. Partial Least Squares based structural equation modeling (PLS-SEM) and fuzzy set Qualitative Comparative Analysis (fsQCA) were used. PLS-SEM results revealed that social features such as user-to-user interaction, homophily, social ties, and social identity, and technical design such as immersive experience and invisibility significantly affect users’ deviant behavior in the metaverse. The fsQCA results provided insights into the multiple causal solutions and configurations. This study is exceptional because it provided decisive results by understanding the deviant behavior of users based on the symmetrical and asymmetrical approach to virtual networks.

Online polarization and identity politics: An analysis of Facebook discourse on Muslim and LGBTQ+ communities in Finland

AbstractThis study employs a mixed‐methods approach to investigate the nature of Facebook posts related to Muslims and LGBTQ+ individuals in Finland, spanning a period of 4 years. Through the use of the CrowdTangle platform, the researchers extracted and analyzed Facebook posts that encompassed predetermined keywords indicative of potential hate speech. The findings underscored divergent patterns of engagement and sentiment toward these two groups, with implications for the different levels of societal acceptance and tolerance exhibited. Posts related to Muslims typically elicited controversy and were often depicted as threats, whereas posts about the LGBTQ+ community generally advocated for inclusivity. However, persistent negative stereotypes about the LGBTQ+ community were also evident. The analysis also brought to light how political parties strategically used these discourses to steer conversations, consolidate their ideological positions, and mobilize their respective supporters. Grounded in the social identity theory, this study sheds light on the complex dynamics of online political discourse, revealing its far‐reaching impacts on societal attitudes, intergroup relations, and formation of group identities. The nuanced understanding derived from these observations suggests that interventions fostering healthier public discussions on social media platforms could contribute significantly to combating societal division, prejudice, and bias. This research underscores the importance of scrutinizing online discourses to address issues of societal cohesion and social acceptance.

A systematic literature review of hate speech identification on Arabic Twitter data: research challenges and future directions

The automatic speech identification in Arabic tweets has generated substantial attention among academics in the fields of text mining and natural language processing (NLP). The quantity of studies done on this subject has experienced significant growth. This study aims to provide an overview of this field by conducting a systematic review of literature that focuses on automatic hate speech identification, particularly in the Arabic language. The goal is to examine the research trends in Arabic hate speech identification and offer guidance to researchers by highlighting the most significant studies published between 2018 and 2023. This systematic study addresses five specific research questions concerning the types of the Arabic language used, hate speech categories, classification techniques, feature engineering techniques, performance metrics, validation methods, existing challenges faced by researchers, and potential future research directions. Through a comprehensive search across nine academic databases, 24 studies that met the predefined inclusion criteria and quality assessment were identified. The review findings revealed the existence of many Arabic linguistic varieties used in hate speech on Twitter, with modern standard Arabic (MSA) being the most prominent. In identification techniques, machine learning categories are the most used technique for Arabic hate speech identification. The result also shows different feature engineering techniques used and indicates that N-gram and CBOW are the most used techniques. F1-score, precision, recall, and accuracy were also identified as the most used performance metric. The review also shows that the most used validation method is the train/test split method. Therefore, the findings of this study can serve as valuable guidance for researchers in enhancing the efficacy of their models in future investigations. Besides, algorithm development, policy rule regulation, community management, and legal and ethical consideration are other real-world applications that can be reaped from this research.

Detecting multilingual hate speech targeting immigrants and women on Twitter

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.

THE CRIMINAL CONFRONTATION FOR CRIMES OF DISCRIMINATION AND HATE SPEECH: A COMPARATIVE STUDY

Background: The crime of disseminating hate speech is considered amongst the most prominent crimes in this era, particularly with the rapid technological advancements witnessed globally, which have contributed to its emergence across all communities. Given its seriousness, this crime threatens societal stability and security. In a bid to foster a culture of global tolerance and encounter the various manifestations of discrimination and racism, the Emirati legislator enshrined criminal protection for individuals against discrimination and hate speech based on ethnicity, race and religion. This legal safeguard is articulated in Federal Law No. 34 of 2023 concerning Combating Discrimination, Hatred and Extremism. Moreover, the Federal Decree-Law No. 34 of 2021, which addresses Combating Rumors and Cybercrimes, further reinforces this legal framework, specifically targeting the spread of such offences through digital channels. This addition underscores the UAE’s comprehensive approach to addressing hate speech and discrimination, acknowledging the evolving nature of these crimes in an increasingly connected world. Methods: The research aims to explore the effectiveness of Emirati law in combating discrimination and hate speech crimes through a multi-method approach, which includes a comparative dimension. Firstly, we conducted a comprehensive literature review of existing legal texts and scholarly articles to understand the broader legal context and historical perspectives on these crimes. This review extended beyond Emirati legal sources to encompass international legal standards and comparative legal analyses, allowing us to contextualise the Emirati legal framework within a global perspective. Results and Conclusions: Our study has provided a detailed examination of the complexities and challenges in addressing the crimes of disseminating and promoting discrimination and hate speech within the UAE legal framework. We have identified that these crimes are inherently complex, requiring a specific intent to provoke violence and discrimination. Our analysis reveals a gap in the current legal approach, particularly in addressing the full scope of hate crimes and the nuances of criminal intent. In light of these findings, we have proposed several critical amendments to the UAE law combating discrimination and hate speech. These include refining the definitions and scope of discrimination in Article 1, incorporating "motive for hate" as a key element in Article 4, revising Article 10 to focus on general criminal intention, and enhancing penalties in Article 16 when foreign financial support is involved. These recommendations aim to strengthen the legal framework, making it more comprehensive and effective in combating discrimination and hate speech, thereby safeguarding social security and human rights. This research contributes significantly to understanding hate speech and discrimination crimes in the UAE and offers practical solutions for legal reform. It underscores the need for continuous evaluation and adaptation of laws to address evolving social challenges effectively.

Empowering hate speech detection: leveraging linguistic richness and deep learning

Social media has become a vital part of most modern human personal life. Twitter is one of the social media that was formed from the development of communication technology. A lot of social media gives users the freedom to express themselves. This facility is misused by users, so hate speech is spread. Designing a system to detect hate speech intelligently is needed. This study uses the hybrid deep learning (HDL) and solo deep learning (SDL) approach with the convolutional neural networks (CNN) and bidirectional gated recurrent unit (Bi-GRU) algorithm. There are 4 models built, namely CNN, Bi-GRU, CNN+Bi-GRU, and Bi-GRU+CNN. Term frequency-inverse document frequency (TF-IDF) is used for feature extraction, which is to get linguistic features to be analyzed and studied. FastText is used to perform feature expansion to minimize mismatched vocabulary. Four scenarios are run. CNN with an accuracy of 87.63%, Bi-GRU produces an accuracy of 87.46%, CNN+Bi-GRU provides an accuracy of 87.47% and Bi-GRU+CNN provides an accuracy of 87.34%. The ability of this approach to understand the context is qualified. HDL outperforms SDL in terms of n-gram type, where HDL can understand sentences broken down by hybrid n-gram types, namely Unigram-Bigram-Trigram which is a complex n-gram hybrid.

Hate speech and cancellation phenomena and the impacts on young people's communication in digital environments.

The ICT Domicílios 2021 survey reveals the growth in the use of the internet in Brazil compared to periods prior to the Covid 19 pandemic. However, this expansion is not reflected in the percentages of selfauthored content posted by users: only 31% reported that they share something of their own creation, be it a text, image, photo, video or music. Among young people from Salvador (Bahia, Brazil) who are interested in politics, Silva's research (2022) revealed that only 34% claim to publish authorial content. Among the reasons given by young people, they mention fear of cancellation and hate speech. This article analyzes data from this survey in the light of discussions on infocommunicative literacy. We conclude that public policies in the country need to advance in terms of guaranteeing the right to the internet associated with education for the ethical, critical and democratic use of information and communication.

Religião e política: embate de sentidos sobre a fé evangélica em posts de pastores no Instagram

ABSTRACT This work aims to reflect on possible meanings for the evangelical faith based on utterances produced by three evangelical ministers in their social networks when they enunciate on themes that are linked, directly or indirectly, to political agendas. Based on the theoretical scope of studies in Applied Linguistics and on Bakhtinian conceptions of language, our analyzes point to three predominant meanings, which mean the evangelical faith as: i) reaffirmation of identity; ii) exercise of otherness and respect to the other; and iii) political practice. Such meanings demonstrate that there are contradictions within the religious discursivity, which allows the beating between hate speeches and respect and empathy ones within the realm of the evangelical faith. Thus, looking for cracks in religious discourse that oppose fundamentalist and authoritarian discourses is an ethical duty of a religious leader who recognizes his social and discursive role as influential in taking positions and, therefore, in the lives of the members of their congregation.

Hate Speech Detection

Abstract: The rise in popularity of microblogging sites such as Facebook, Instagram, and Twitter has resulted in more people from different backgrounds indirectly communicating with one another. Our study aims to design an autonomous Deep Neural Network (DNN) algorithm for social media hate speech detection to tackle this problem. Using cutting-edge Natural Language Processing (NLP) techniques, the objective is to build a strong system that can recognize and categorize hate speech material in text data with accuracy. Our DNN algorithm allows for the real-time detection and moderation of offensive information, providing a proactive strategy against online hate speech. With the deployment of this technology, everyone will be able to access a safer and more welcoming online environment.