Identification of Hate Speech in Social Media

Social media platforms provide an inexpensive communication medium that allows anyone to quickly reach millions of users. Consequently, in these platforms anyone can publish content and anyone interested in the content can obtain it, representing a transformative revolution in our society. However, this same potential of social media systems brings together an important challenge---these systems provide space for discourses that are harmful to certain groups of people. This challenge manifests itself with a number of variations, including bullying, offensive content, and hate speech. Specifically, authorities of many countries today are rapidly recognizing hate speech as a serious problem, specially because it is hard to create barriers on the Internet to prevent the dissemination of hate across countries or minorities. In this paper, we provide the first of a kind systematic large scale measurement and analysis study of hate speech in online social media. We aim to understand the abundance of hate speech in online social media, the most common hate expressions, the effect of anonymity on hate speech and the most hated groups across regions. In order to achieve our objectives, we gather traces from two social media systems: Whisper and Twitter. We then develop and validate a methodology to identify hate speech on both of these systems. Our results identify hate speech forms and unveil a set of important patterns, providing not only a broader understanding of online hate speech, but also offering directions for detection and prevention approaches.

Detecting hate speech in social media is challenging due to its rarity, high-dimensional complexity, and implicit expression via sarcasm or spelling variations, rendering linear models ineffective. In this study, the SVM (Support Vector Machine) algorithm is used to map text features from low-dimensional to high-dimensional space using kernel function techniques to meet complex nonlinear classification challenges. By maximizing the category interval to locate the optimal hyperplane and combining nuclear techniques to implicitly adjust the data distribution, the classification accuracy of hate speech detection is significantly improved. Data collection leverages social media APIs (Application Programming Interface) and customized crawlers with OAuth2.0 authentication and keyword filtering, ensuring relevance. Regular expressions validate data integrity, followed by preprocessing steps such as denoising, stop-word removal, and spelling correction. Word embeddings are generated using Word2Vec’s Skip-gram model, combined with TF-IDF (Term Frequency–Inverse Document Frequency) weighting to capture contextual semantics. A multi-level feature extraction framework integrates sentiment analysis via lexicon-based methods and BERT for advanced sentiment recognition. Experimental evaluations on two datasets demonstrate the SVM model’s effectiveness, achieving accuracies of 90.42% and 92.84%, recall rates of 88.06% and 90.79%, and average inference times of 3.71 ms and 2.96 ms. These results highlight the model’s ability to detect implicit hate speech accurately and efficiently, supporting real-time monitoring. This research contributes to creating a safer online environment by advancing hate speech detection methodologies.

Hate speech in social media is a growing phenomenon, and detecting such toxic content has recently gained significant traction in the research community. Existing studies have explored fine-tuning language models (LMs) to perform hate speech detection, and these solutions have yielded significant performance. However, most of these studies are limited to detecting hate speech only in English, neglecting the bulk of hateful content that is generated in other languages, particularly in low-resource languages. Developing a classifier that captures hate speech and nuances in a low-resource language with limited data is extremely challenging. To fill the research gap, we propose HateMAML, a model-agnostic meta-learning (MAML)-based framework that effectively performs hate speech detection in low-resource languages. HateMAML utilizes a self-supervision strategy to overcome the limitation of data scarcity and produces better LM initialization for fast adaptation to an unseen target language (i.e., cross-lingual transfer) or other hate speech datasets (i.e., domain generalization). Extensive experiments are conducted on five datasets across eight different low-resource languages. The results show that HateMAML outperforms the state-of-the-art baselines by more than 3% in the cross-domain multilingual transfer setting. We also conduct ablation studies to analyze the characteristics of HateMAML.

ABSTRACTIn this study, we approach the problem of distinguishing general profanity from hate speech in social media, something which has not been widely considered. Using a new dataset annotated specifically for this task, we employ supervised classification along with a set of features that includes -grams, skip-grams and clustering-based word representations. We apply approaches based on single classifiers as well as more advanced ensemble classifiers and stacked generalisation, achieving the best result of accuracy for this 3-class classification task. Analysis of the results reveals that discriminating hate speech and profanity is not a simple task, which may require features that capture a deeper understanding of the text not always possible with surface -grams. The variability of gold labels in the annotated data, due to differences in the subjective adjudications of the annotators, is also an issue. Other directions for future work are discussed.

Commonly used language defines "hate speech" as objectionable statements that may jeopardize societal harmony by singling out a group or a person based on fundamental traits (including gender, caste, or religion).Using machine learning techniques, our research focuses on identifying hate speech in social media comments.Using a variety of machine learning methods, we created machine learning models to detect hate speech.An approximate Macro F1 of 0.60 was attained by the created models.

As the use of internet and social media increases, hate speech has emerged as a significant issue. Hateful remarks are made anonymously by most users, and the incidence of harmful conversations, particularly on controversial subjects, increases. These kinds of content make readers feel upset and help create an unhealthy online environment. This study aims to address this problem by exploring the use of transformer models in hate speech identification and moderation of Malayalam social media posts. Based on a dataset from the DravidianLangTech competition, we train and test various transformer models for hate speech classification. This study aims to improve automated hate speech detection and removal tools for social media to provide a cleaner digital environment.

This article investigates the identification of hate speech on social media using machine learning and deep learning techniques. The research uses metrics such as F-measure, AUC-ROC, precision, accuracy, and recall assessing the effectiveness of various tactics. The findings indicate that deep learning models, particularly the bidirectional long short-term memory (BiLSTM) architecture, consistently outperform other methods in categorization tasks. The research emphasizes the importance of sophisticated neural network designs in identifying the intricacies of hostile and offensive content online. The study offers insights for promoting early identification and prevention of cyberbullying, improving secure and inclusive online environments. Future research may explore real-time detection systems, hybrid approaches, or the integration of complementary components to enhance and improve innovative technology in tackling this significant social issue. A sample tweet was annotated by specialists who categorize tweets as hate speech, offensive language, or neutral. The researchers applied shallow learning methodologies and integrated word embeddings like Word2Vec and GloVe to enhance the efficacy of deep learning models. The results indicate that BiLSTM surpasses shallow learning methods in detecting hate speech on Twitter, highlighting the efficacy of deep learning approaches in recognizing and tracking hate speech on social media platforms. When comparing different deep learning and machine learning models on different datasets, the results reveal that deep learning techniques are usually more effective. A reasonably high level of accuracy is achieved by KNN and SVM among classical algorithms, whereas Naïve Bayes performs the poorest. While deep learning approaches provide better results, tree-based models such as Random Forest and Decision Trees offer more consistent accuracy. Models based on neural networks, such as LSTM, CNN, and BI-LSTM, perform well, with LSTM-based methods excelling in particular. The most successful strategy for classification problems is the model presented, which obtains the greatest accuracy, precision, recall, F1-score of 95%. The research aids in the development of advanced tools and methodologies to mitigate hate speech on social media and foster positive online interactions. Future research may investigate alternative deep learning architectures, such as transformers, to enhance hate speech detection efficacy. The advancement of interpretable AI methodologies for identifying hate speech and delivering transparent forecasts might enhance user confidence and facilitate better content moderation decisions.

It is critical to have efficient automated detection systems in place since the proliferation of hate speech and inflammatory language on social media platforms has been accelerated by their fast development. Attempts to capture contextual connections were a challenge for traditional machine learning methods like SVM and Logistic Regression, which depended on manually created features. Though they mostly handled text as sequences and neglected underlying relationship structures, deep learning and Transformer-based models learned contextual embeddings, which enhanced performance. To overcome this shortcoming, this research presents a GNN-based framework for modeling textual data as graphs, which allows for a more comprehensive portrayal of the semantic and relational relationships between words and texts. For experimental assessment, three datasets were used: Davidson, which had 24,783 tweets with 5.8% hate, 77.4% offensive, and 16.8% neither; HASOC, which contained around 9,000-12,000 tweets in English and Hindi, with a somewhat balanced distribution; and Founta, which contained 80,000+ tweets in all three categories (hate, abusive, and normal). Baselines such as Logistic Regression, SVM, CNN, BiLSTM, and BERT classifiers were tested against the suggested architecture, which combines GCN, GAT, and GraphSAGE with pre-trained embeddings (GloVe and BERT). Dataset after dataset shows that GNN-based models perform far better than baselines. The F1-scores of GraphSAGE and BERT on the Davidson dataset were 91.7% and 87.8%, respectively. For the HASOC dataset, GraphSAGE achieved an F1-score of 89.9%, which was higher than BERT's 85.5%. Again, GraphSAGE topped the Founta dataset with a 90.2% F1-score, well surpassing BERT's 86.1% performance. Compared to the most advanced Transformer models, these findings show an improvement of 3-5% in F1-score. The results show that abusive language and hate speech identification are better handled by graph-based representations, which give a more complete picture of the linguistic and relational environment. This study lays the groundwork for future research on multimodal, multilingual, and real-time detection systems, and it also sets GNNs as a strong and scalable method for online content moderation. Keywords: Handwritten Hate Speech, Offensive Language, Graph Neural Networks (GNNs), GCN, GAT, GraphSAGE, Social Media, Text Classification,

The internet has become an everyday tool to communicate and network with people around the globe, but its perceived anonymity, availability, and instant access have made it an environment conducive to spreading hateful content and connecting to like-minded individuals with similar hateful ideologies. Hate speech and other prejudice-motivated behavior, however, need to be considered on a continuum of victimization, and "like other social processes, [be seen as] dynamic and in a state of constant movement and change, rather than static and fixed" (Bowling, 1993, p. 238). It is a social process that is marked by multiple, repeat, and constant victimization (Bowling, 1993), with victims no longer distinguishing between specific hateful events, and rather normalizing experiences of hateful conduct "as an everyday, unwanted but routine reality of being 'different'" (Chakraborti, 2016, p. 581). Understanding hateful behavior and victimization as a process allows us to connect "low-level" incidents of hateful behavior to the more serious and life-threatening incidents at the more extreme end of the spectrum (Bowling & Phillips, 2002). The Christchurch attacks in New Zealand and their link to hateful communication on the online platform 8chan is only one such example of how online hate speech and cyberhate can escalate to "in real life" attacks, leaving the online sphere and spilling into the offline world. As per Allport's (1954) scale of prejudice, more extreme forms of prejudice-motivated violence are founded on "lower level" acts of prejudice and bias, therefore, hateful content online should not be ignored. Intervening online to interrupt or counter hateful behavior already at the lower end of the scale of prejudice becomes important; online interventions which are to be identified and synthesized through this systematic review. Allport's (1954) scale of prejudice will be the basis for this systematic review. Early on, Allport (1954) asserted that individuals with negative attitudes toward groups are likely to act out on these prejudices "somehow, somewhere" (p. 14), and that the more intense such negative attitudes are, the more hostile the action will be. Allport (1954) put forward a scale of acts of prejudice to illustrate different degrees of acting out negative attitudes, a scale that starts with antilocution (or what we call hate speech), described as explicitly expressing prejudices through negative verbal remarks to either friends or strangers (Allport, 1954). Avoidance is the next level on the scale of prejudice, with people avoiding members of certain groups, followed by discrimination, where distinctions are made between people based on prejudices, which leads to the active exclusion of members from certain groups (Allport, 1954). This level of acting on prejudices is routed in institutional or systemic prejudices, for example, in the differential treatment of people within employment or education practices, but also within the criminal justice system, or through social exclusion of certain minority group members. Physical attack is the next level on the scale of prejudice, which includes violence against members of certain groups by physically acting on negative attitudes or prejudices. The last level is extermination, which is the ultimate act of violence against members of specific groups, an expression of prejudice that systematically eradicates an entire group of people (e.g., genocide or lynchings; Allport, 1954). Allport's (1954) scale of prejudice makes it clear how hate speech/cyberhate is connected to more extreme forms of violence motivated by specific prejudices and biases, with hate speech (or antilocutions) being only the starting point on a 5-point continuum (Bilewicz & Soral, 2020). The importance of this scale of prejudice is not only that it clearly illustrates a range of different ways and intensity levels to act out prejudices, but also the "progression from verbal aggression to physical violence or, in other words, the performative potential of hate speech" (Allport, 1954; Kopytowska & Baider, 2017, p. 138). This is where interventions at the lower level of the scale of prejudices, interventions targeting hate speech/cyberhate, become important. There is no universal definition of hateful conduct online, but there is some consensus that hate speech targets disadvantaged social groups (Jacobs & Potter, 1998). Bakalis (2018) more narrowly defines cyberhate as "any use of technology to express hatred towards a person or persons because of a protected characteristic—namely race, religion, gender, sexual orientation, disability and transgender identity" (p. 87). Another definition that also points out the ambiguity and challenges involved with identifying more subtle forms of hate speech, and also making reference to the potential threat of hate speech escalating to offline violence, is that put forward by Fortuna and Nunes (2018), who analyzed various definitions of hate speech "Hate speech is language that attacks or diminishes, that incites violence or hate against groups, based on specific characteristics such as physical appearance, religion, descent, national or ethnic origin, sexual orientation, gender identity or other, and it can occur with different linguistic styles, even in subtle forms or when humour is used" (p. 5). In this systematic review, we also distinguish hate speech/cyberhate specifically from other forms of harmful online activity, such as cyber-bullying, harassment, trolling or flaming, as perpetrators of such online behavior repeatedly and systematically target specific individuals to cause upset, to seek out negative reactions, or to create discord on the internet. In contrast, hate speech/cyberhate is more general and does not necessarily target a specific individual (Al-Hassan & Al-Dossari, 2019), instead hate speech/cyberhate heavily features prejudice, bias and intolerance toward certain groups within society. With the majority of hate speech happening online, interventions that take place online are an important way to challenge prejudice and bias, potentially reaching masses of people across the globe. The unique feature of the internet is that such individual negative attitudes toward minority groups and more extreme hateful ideology can find its way onto certain platforms and can instantly connect people sharing similar prejudices. By closing the social and spatial distance, the internet creates a form of collective identity (Perry, 2000, p. 123) and can convince individuals with even the most extreme ideologies that others out there share their views (Gerstenfeld et al., 2003). In addition, the enormous frequency of hate speech/cyberhate within online environments creates a sense of normativity to hatred and the potential for acts of intergroup violence or political radicalization (Bilewicz & Soral, 2020, p. 9). It is, therefore, important to challenge this hate speech epidemic (Bilewicz & Soral, 2020), especially since hate movements have increasingly crossed into the mainstream (Perry, 2000). With hate speech/cyberhate posing a threat to the social order by violating social norms (Soral et al., 2018), perceptions of social norms as either supporting or opposing prejudice has been found to have an influence on how individuals react online (Hsueh et al., 2015). Seeing other people post prejudiced (opposed to antiprejudiced) comments online can lead to the adoption of an online group's biases and can influence an individual's own perceptions and feelings toward the targeted stigmatized group (Hsueh et al., 2015). In addition, research around desensitization also suggests that being exposed to hate speech leads to desensitization, which further leads to an increase in outgroup prejudice toward groups targeted by such speech (Soral et al., 2018). With society increasingly recognizing that it is inappropriate to express prejudices in public settings, many interventions will include some form of social norms nudging to reduce such prejudices; interventions that "nudge behavior in the desired direction" (Titley et al., 2014, p. 60). Therefore, hate speech not only affects minority group members, but also has an influence on opinions of majority group members (Soral et al., 2018), which makes strategies that can elicit change in people's prejudice-related attitudes crucial (see, e.g., Zitek & Hebl, 2007). Governments around the world face increased demand for understanding and countering hateful ideology and violent extremism both online and offline (e.g., the Christchurch Call in New Zealand). The U.S. Government's 2011 CVE Strategy highlights the importance of ongoing research and analysis, the sharing of knowledge and best practices internationally, and the countering of hateful ideologies and propaganda (see also Department of Homeland Security, 2016, 2019). The goal of this systematic review is to use an integrated and interdisciplinary approach to examine the effectiveness of online campaigns and strategies for reducing hate speech and cyberhate. The internet also provides an opportunity to reach masses of people who have been exposed to hateful content and ideology online, therefore, this systematic review will focus on online interventions addressing online hate speech and cyberhate. The specific settings where we would expect to see the online interventions deployed will be on websites, text messaging applications, and online and social media platforms including, but not limited to, Facebook, Instagram, TikTok, WhatsApp, Google, YouTube, and Snapchat. As mentioned previously, many online interventions will be based on social norm nudges to reduce online hate. These interventions aim to change people's online behavior and encourage individuals or groups to conform to established social norms. The communication of social norms can happen through establishing community standards on online platforms themselves (e.g., Facebook, Twitter, etc.), through more formal online training courses, or through anti-hate speech/anti-cyberhate campaigns teaching people to recognize hate, embrace diversity, and stand up to bias. Such prevention campaigns are designed to challenge bias and build ally behaviors by supplying people with constructive responses to combat, for example, antisemitism racism, and homophobia, as well as provide resources to help people explore and critically reflect on current events. Other interventions may add messages to hateful online comments, counter hateful content or extremist ideology, or redirect people to more credible sources. Both peers and parents have been found to foster racial consciousness and identity development, define interracial relationships and cultivate ethnic heritage and culture (Hagerman, 2016). Socialization influences how children understand their group's social position and their membership within that group by providing an understanding of racial, religious, and sexual privilege (Bowman & Howard, 1985). Socialization often reflects peers' and parents' experiences with racism, discrimination, and their ideological perspectives about race, religion, or sexuality (Umaña-Taylor & Fine, 2004). This is important because peers and parents who feel discriminated against or believe that the "other" is a threat may impart their prejudices to their children or friends, which could lead them to interpret the social world with similar discriminatory views and/or behavior. Individuals who feel socially alienated or rejected are especially vulnerable to such socialization practices as they feel that adopting these views will provide them with a sense of acceptance and belonging (Leiken, 2012). Regardless of how an individual develops certain racial, religious, or sexual biases, the online interventions under review are expected to target and reduce the production of original hateful content such as antisemitic Tweets and/or homophobic blog posts as well as the consumption of hate speech material (e.g., watching or reading hate speech videos or blogs). For example, some interventions take a rather broad messaging approach by implementing racial sensitivity and diversity training through Public Service Announcements, peer-to-peer dialogue workshops, or films that provide opportunities for youth and adults to self-reflect and learn about historical oppression, people of color, women, and the LGBTQIA+ community from credible sources. The factual understanding of diverse groups is often supplemented by experiences with people within the group. These educational programs often identify a cultural guide who is willing to introduce youth to new experiences and who can aid in processing thoughts, feelings, and behaviors. These interventions intend to dispute and contradict negative stereotypes associated with specific cultures, people, and institutions by sharing different points of view based on human rights values such as openness, respect for difference, freedom, and equality (Gomes, 2017). Moreover, such interventions tend to involve blanket bans on specific behaviors enforced through the public promotion of norms or individual sanctions enforced by moderators. Other interventions, such as the "Redirect Method," are narrower in their messaging. These interventions generate curated playlists and collections of authentic content that challenge hate speech/cyberhate narratives and propaganda (Helmus & Klein, 2018). For instance, people who are directly searching for extremist content online may be linked to videos and written content that confronts such claims. These videos are designed to be objective in appearance instead of containing material that explicitly counters extremist propaganda. The underlying goal of this type of interventions is to provide credible content that effectively undermines extremist messaging but does not overtly attack the source of propaganda. In addition to confronting hate speech narratives, these interventions provide users with links to numerous social services such as anger management training, drug and alcohol treatment, and mental health resources. Online platforms, such as Twitter and Facebook, have also started to employ a similar method, redirecting people who comment on or share "fake news" or conspiracy theories, which often are fraught with prejudicial undertones and are harmful to minority groups, to more credible content and news sources. The aforementioned interventions are designed to counter-balance these biased perceptions (e.g., unsupported claims of the Black community as criminal or the LGBTQIA+ community as pathologized) Blacks as criminals, LGBTQIA+ as pathologized) by blunting the occurrence of racist discourse and reducing the likelihood these individuals will internalize and normalize racial, religious, and/or sexual prejudices (Qian et al., 2019). Being in new situations is uncomfortable and often awakens fears and apprehensions that can block our experiential development. Acquiring information or being exposed to minority-run businesses, poverty, and writings from minority authors allows a person to understand the thoughts, hopes, fears, and aspirations of the people outside their racial perspective rather than from the perspective of the majority society (Dunham et al., 2013; Lee et al., 2017). Doing so, counters racist programming by challenging hegemonic beliefs, which can lead to the acceptance of tolerant attitudes and the reduction of hateful expressions online. Findings from the proposed review will enhance our understanding of the effectiveness of online anti-hate speech/anti-hate interventions, will help ensure that programming funds are dedicated to the most-effective efforts, and will play a critical role in helping individual programs improve the quality of service provisions. It will inform governments and policymakers of the current state of such online efforts, what works and which modes of interventions to implement, and help guide economically viable investments in nation-state security. Our search of the scholarly literature identified one review, Blaya (2019), as similar to the proposed topic. Blaya's (2019) review, however, focused on the prevalence, type, and characteristics of existing interventions for counteracting cyberhate and did not include a meta-analysis. Two other similar reviews focused on exposure to extremist online content (Hassan et al., 2018) and communication channels associated with cyber-racism (Bliuc et al., 2018). A search of the Campbell Library using key terms (hate OR radical*) returned two protocols and one review identified for further inspection to assess potential overlap. The protocols include "Psychosocial processes and intervention strategies behind Islamist deradicalization: A scoping review" by de Carvalho et al. (2019) and "Police programs that seek to increase community connectedness for reducing violent extremism behavior, attitudes and beliefs" by Mazerolle et al. (2020). A further review on a similar topic is a recently completed Campbell review (January 2020), "Counter-narratives for the prevention of violent radicalization: A systematic review of targeted interventions" by Carthy et al. (2018) at the National University of Ireland, Galway. Our proposed review is distinguished from the de Carvalho et al. (2019) review in that we are focusing on hate speech and cyberhate generally without delimiting our approach to a specific type of radicalization (e.g., Islamist). Furthermore, we are electing to complete a systematic review and meta-analysis. Likewise, the protocol by Mazerolle et al. (2020) focuses on interventions involving police officers either as initiators, recipients, or implementers of community connectedness interventions. Our review will focus specifically on any online intervention, which may or may not involve police, but police will not be the focus nor be the basis of the online intervention strategy. Judging from Carthy et al. (2018) protocol, we anticipate our review will also capture counter-narrative interventions, but will differ based on setting, timing, and scope of interventions. Specifically, we are interested in online interventions that extend beyond counter-messaging campaigns to include a broad array of interventions outlined above and extend beyond radicalization to include everyday hate and prejudice. In addition to conducting a meta-analysis, the proposed review would build on Blaya's (2019) work by expanding the population parameters to include both adolescents as well as adults. Blaya (2019) limited her search to include interventions aimed toward youth, young people, children, young adults, adolescents, children, and teenagers and did not focus on extremism. The main objective of this review is to synthesize the available evidence on the effectiveness of online interventions aimed at reducing the creation and/or consumption of online hate speech/cyberhate material. To what extent are online interventions effective in reducing online hate speech/cyberhate? How is effectiveness related to the type of online hate speech/cyberhate intervention used? How is effectiveness related to the characteristics of individuals experiencing the online hate speech/cyberhate intervention (e.g., age, gender, race/ethnicity, offense history, childhood trauma)? Both experimental and quasi-experimental quantitative studies will be included. These study designs will address research questions #1 to #3. Eligible quantitative study designs include the following: Eligible experimental designs must involve random assignment of participants to distinct treatment and control group(s). Designs that involve quasi-random assignment of participants such as alternate case assignment are also eligible and will be coded as experimental designs. All eligible quasi-experimental designs must include a comparison group of participants compared to participants in the treatment condition. Eligible studies include those that report matching procedures (individual- or group-level) and statistical procedures employed to achieve equivalency between groups. Statistical procedures may but are not limited to, analysis, and Furthermore, in of a limited quantitative evidence we will also include quasi-experimental studies with comparison groups that provide of for both groups. will also be included. Eligible include designs with a control group and designs with or without a control group than quasi-experimental designs include studies that a comparison group of participants who either to in the study or who in a but out to the of a Eligible comparison include other online interventions or in which participants not or an online Both youth and participants of any gender sexual orientation, or will be eligible for this review. The eligible youth population will be study participants with a of through The eligible population will be study participants with a of and in which only a of the is eligible for example, a study in both online and offline hate speech be not anticipate studies based on as our will be and we will take to studies that only online interventions. will of the of a study for through and be we will elicit the of a the of and studies will be these studies will be they will be and be in the and any related Blaya's (2019) of intervention strategies to the potential of eligible interventions. The intervention is the of responses to hate speech/cyberhate, which includes the countering of violent extremism and to address online interventions that are eligible range from hateful content online specific (e.g., of social media to to online hate using targeted strategies (e.g., through hateful of studies focusing on online include the and of online and content online content and & 2018), hateful online comments to comments et al., 2018), and to users out of online are also interested in interventions such as the of 8chan this online platform linked to "in real life" attacks in New Zealand and the and interventions that further hateful online content and radicalization similar events. hateful content online such has up speech as well as around online users and hateful groups on to other online to hateful content online using targeted strategies therefore, been as an effective online include using the from & 2020), the use of to online responses to in online where hate speech has been (Qian et al., 2019), and redirecting online users to videos for example, Our systematic review will include a range of online interventions, many of which have only recently Two other strategies identified by Blaya (2019) are the and of hate speech/cyberhate using technology as well as the creation of online and These interventions include online counter-narrative the and/or use of online counter online interventions, online training, and online narrowly to address extremist ideologies and hate speech that incites targeted violence and In such interventions seek to or the occurrence of violent extremism or the of hate speech and extremist by channels and opportunities to such groups. The and intervention eligible for this systematic review educational programs for example, provide people with online and challenge 2019). will include online programs with an online (e.g., and and educational and online interventions. of these interventions may by individuals no longer in the creation and/or consumption of cyberhate and extremist material online. These online interventions may be by and internet service or or in the case of interventions. The comparison may be routine exposure and to hate speech/cyberhate or online The of is the creation and/or consumption of hateful content online. By we to the production and of original hateful content such as antisemitic racist and/or homophobic blog The consumption of hate speech material may include or being a of a hate watching or reading hate speech videos or being a target of online hate speech/cyberhate, or hate speech material. of include and of study participants such as and attitudes toward hate Eligible studies must report a or (or to be included. There will be no exclusion on the source of for the and can be from any institutional or completed by will include any of from strategies to increase the scale of of potentially effective anti-hate speech and interventions for These could include to or to the creation of and behaviors. can also include such as a of hate speech/cyberhate to other platforms instead of a reduction of hate All described in eligible studies will be in the will focus on the between and the current The starting with the when the internet to a and community et al., are for an approach in the lower end of our search to the may be it is hate speech/cyberhate online through or and some studies may capture Our population of studies will also be limited to studies in and but of studies completed in any as we are focused on online content that can be and across and nation-state The language parameters reflect the language of the review Our will where studies the of study in will be between the members of the review These will be and as a from the protocol in the review. In the of a change in we will search online OR OR internet OR Twitter OR OR 8chan OR OR OR OR OR OR OR OR OR OR speech" OR cyberhate OR OR OR OR OR speech OR OR OR OR OR OR OR OR OR OR OR OR OR OR OR OR OR OR OR peer-to-peer OR OR OR

Societal and technological challenges are significant when it comes to cyberbullying, which is an ubiquitous issue in the social media space including Twitter, Facebook, YouTube, and Instagram. This paper will focus on the identification of hate speech that targets particular individuals, especially in the context of the data in Hindi language. It tries to fill the gap between the overall abusive language and the hatred directed at specific people or populations. To this end, an annotated and carefully edited compilation was created where hate speech was divided into the following categories: racial/ethnic, religious, sexual orientation, and political. In order to address the problem of the imbalance of the classes, both Synthetic Minority Over-sampling Technique (SMOTE) and Grouped SMOTE were applied to the model to enhance its efficiency. The traditional machine learning frameworks (Support Vector Machines (SVM), Convolutional Neural Networks (CNN), Long Short-Term Memory networks (LSTM), and the transformer-based model Bidirectional Encoder Representations from Transformers (BERT) have been explored and the results show that BERT outperforms the traditional models (the F1-score of BERT is 89.2), which supports the initial hypothesis. The second hypothesis was also supported by the use of SMOTE which increased accuracy and precision. It was found that there was a strong correlation between the frequency of hate speech and the different demographic attributes particularly with regard to racial and political biases which support the third hypothesis. Also, the document underlines the fact that goal-based classification is a better approach to binary classification models, thus validating the fourth hypothesis. Further analysis reveals a difference in hate speech trends on different platforms: on Twitter, politically charged hate is quite widespread, and on Facebook, hate speech is mostly based on religious topics. The outlined findings emphasise the importance of creating detection systems of hate speech that are platform-adaptable, demographically aware, and that are also built upon sentiment analysis. This study contributes to the field of context-sensitive content control and enhances fairness of hate speech detection through natural languages processing tools and is of great interests to the researcher of artificial intelligence, decision-makers, and social networks.

Hate speech in social media nowadays is a common thing to happen. Inspired by the issue, this research utilize data mining algorithm and methods to predict and classify it. By using dataset from twitter, this research will focus to define Hate Speech. Before beginning to use the algorithm, firstly the dataset needs to be cleaned, after that the data will be converted tu numeric values by using TF-IDF. With N-Gram, the final results will be more stable in terms of accuracy. After the preprocessing is done, then the K-Means Algorithm is used. The final results of the research is that by using Tri-Gram, accuracy is better than Bi-Gram and Uni-Gram with highest reach of 80% efficiency.

In recent times, the Internet and social media are very well-known and popular among people. Usage of social media is increased exponentially during the last few years globally, and it allows people to engage with one another and share ideas, thoughts, opinions, etc. Every day, massive amounts of data are disseminated at breakneck speed via social media platforms, reaching a massive audience. Furthermore, the ability to write anonymous posts and comments makes expressing and spreading hate speech even easier. To improve the users’ experience, social media sites are attempting to remove hateful remarks. In this research work, the main focus is on developing automated hate speech and offensive language detection models. Started with traditional hate speech and offensive language identification approaches, than reaches to advanced hate speech recognition methods for social media. This brings a need for integrated datasets and the hate speech prediction method. This paper describes the study on hate speech and offensive content identification in English language by using the various approaches based on machine learning algorithms (Support vector machine, decision tree, and so on) and NLP, along with the features used for the classification problem. The models were tested on the HASOC (2021) datasets and concluded that the ensemble model perform better than other algorithms with the test dataset for different tasks. Results and analysis part of this paper offers researchers a comprehensive picture of approaches.

This paper provides an overview of state-of-the-art approaches toward the identification of hate speech through Natural Language Processing and Machine Learning methods. These models that have been adopted include Bi-LSTM, CNN, and GPT-2, highlighting the merits and demerits. The best detection accuracy for hate speech was obtained using the model GPT-2 with a 97.12% rate because it understands the context as well as generates words in sequences. More profound challenges the paper deals with include linguistic ambiguity, dataset bias, and scalability. New contributions involve developing preprocessing pipelines from scratch, hybrids between models, and real-time deployment techniques. The end includes the ethical implication of the deployed automated hate speech detection systems and directions for future work in generating more culturally adaptable, multilingual, and computationally efficient models.

Online hate has emerged as a rapidly growing issue worldwide, often stemming from differences in opinion. It is crucial to use appropriate language and words on social media platforms, as inappropriate communication can negatively impact others. Consequently, detecting hate speech is of significant importance. While manual methods are commonly employed to identify hate and offensive content on social media, they are time-consuming, labor-intensive, and prone to errors. Therefore, AI-based approaches are increasingly being adopted for the effective classification of hate and offensive speech. The proposed model incorporates various text preprocessing techniques, such as removing extraneous elements like URLs, emojis, and blank spaces. Following preprocessing, tokenization is applied to break down the text into smaller components or tokens. The tokenization technique utilized in this study is TF-IDF (Term Frequency–Inverse Document Frequency). After tokenization, the model performs the classification of hate and offensive speech using the proposed BiLSTM-based SM-CJ (Scalable Multi-Channel Joint) framework. The BiLSTM-based SM-CJ model is particularly effective in detecting hate, offensive, and neutral tweets due to its ability to capture both forward and backward contexts within a given text. Detecting hate speech requires a comprehensive understanding of the text and the identification of patterns spanning across multiple words or phrases. To achieve this, the LSTM component of the BiLSTM model is designed to capture long-term dependencies by utilizing information from earlier parts of the text. The proposed SM-CJ framework further aligns the input sequence lengths fetched from the input layer, enabling the model to focus on specific segments of the input sequence that are most relevant for hate speech detection. This approach allows the model to accurately capture derogatory language, and subtle nuances present in hate speech. Finally, the performance of the proposed framework is evaluated using various metrics, including accuracy, recall, F1-score, and precision. The results are compared with state-of-the-art approaches, demonstrating the effectiveness of the proposed model. Doi: 10.28991/ESJ-2025-09-01-03 Full Text: PDF

PurposeHate speech in recent times has become a troubling development. It has different meanings to different people in different cultures. The anonymity and ubiquity of the social media provides a breeding ground for hate speech and makes combating it seems like a lost battle. However, what may constitute a hate speech in a cultural or religious neutral society may not be perceived as such in a polarized multi-cultural and multi-religious society like Nigeria. Defining hate speech, therefore, may be contextual. Hate speech in Nigeria may be perceived along ethnic, religious and political boundaries. The purpose of this paper is to check for the presence of hate speech in social media platforms like Twitter, and to what degree is hate speech permissible, if available? It also intends to find out what monitoring mechanisms the social media platforms like Facebook and Twitter have put in place to combat hate speech. Lexalytics is a term coined by the authors from the words lexical analytics for the purpose of opinion mining unstructured texts like tweets.Design/methodology/approachThis research developed a Python software called polarized opinions sentiment analyzer (POSA), adopting an ego social network analytics technique in which an individual’s behavior is mined and described. POSA uses a customized Python N-Gram dictionary of local context-based terms that may be considered as hate terms. It then applied the Twitter API to stream tweets from popular and trending Nigerian Twitter handles in politics, ethnicity, religion, social activism, racism, etc., and filtered the tweets against the custom dictionary using unsupervised classification of the texts as either positive or negative sentiments. The outcome is visualized using tables, pie charts and word clouds. A similar implementation was also carried out using R-Studio codes and both results are compared and a t-test was applied to determine if there was a significant difference in the results. The research methodology can be classified as both qualitative and quantitative. Qualitative in terms of data classification, and quantitative in terms of being able to identify the results as either negative or positive from the computation of text to vector.FindingsThe findings from two sets of experiments on POSA and R are as follows: in the first experiment, the POSA software found that the Twitter handles analyzed contained between 33 and 55 percent hate contents, while the R results show hate contents ranging from 38 to 62 percent. Performing a t-test on both positive and negative scores for both POSA and R-studio, results reveal p-values of 0.389 and 0.289, respectively, on an α value of 0.05, implying that there is no significant difference in the results from POSA and R. During the second experiment performed on 11 local handles with 1,207 tweets, the authors deduce as follows: that the percentage of hate contents classified by POSA is 40 percent, while the percentage of hate contents classified by R is 51 percent. That the accuracy of hate speech classification predicted by POSA is 87 percent, while free speech is 86 percent. And the accuracy of hate speech classification predicted by R is 65 percent, while free speech is 74 percent. This study reveals that neither Twitter nor Facebook has an automated monitoring system for hate speech, and no benchmark is set to decide the level of hate contents allowed in a text. The monitoring is rather done by humans whose assessment is usually subjective and sometimes inconsistent.Research limitations/implicationsThis study establishes the fact that hate speech is on the increase on social media. It also shows that hate mongers can actually be pinned down, with the contents of their messages. The POSA system can be used as a plug-in by Twitter to detect and stop hate speech on its platform. The study was limited to public Twitter handles only. N-grams are effective features for word-sense disambiguation, but when using N-grams, the feature vector could take on enormous proportions and in turn increasing sparsity of the feature vectors.Practical implicationsThe findings of this study show that if urgent measures are not taken to combat hate speech there could be dare consequences, especially in highly polarized societies that are always heated up along religious and ethnic sentiments. On daily basis tempers are flaring in the social media over comments made by participants. This study has also demonstrated that it is possible to implement a technology that can track and terminate hate speech in a micro-blog like Twitter. This can also be extended to other social media platforms.Social implicationsThis study will help to promote a more positive society, ensuring the social media is positively utilized to the benefit of mankind.Originality/valueThe findings can be used by social media companies to monitor user behaviors, and pin hate crimes to specific persons. Governments and law enforcement bodies can also use the POSA application to track down hate peddlers.