Evaluating Semantic Feature Representations to Efficiently Detect Hate Intent on Social Media

Senarath, Yasas; Purohit, Hemant

doi:10.1109/icsc.2020.00041

Cited by 18 publications

(23 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moreover, additional hostile keywords should be acquired to monitor potential hateful texts in response to changes in popular hate themes [31]. A further course of work in the future should include exploring different machine-learning techniques and methods to characterize and track social media user-centered content [52,28,35,94,95,4]. A creative model for increasing hate speech data should be established using data created through a deep generative strategy trained on limited datasets of hate speech [70].…”

Section: Discussion and Future Workmentioning

confidence: 99%

“…The offered system relies on feature selection methods, namely, information gain and term frequency-inverse document frequency. Additionally, [95] provided a detailed evaluation of the importance of different semantic feature representations of social media posts. Semantic features feature can support and enhance the contextual interpretation of the word senses of a machine learning model.…”

Section: E) Separation Of Hate Speech From An Offensive Instancementioning

confidence: 99%

“…[35] established other benchmark datasets and introduced convolutional and gated recurrent networks with six more datasets. Understanding the nuances of messages from the user with the possible intention of hate led to the introduction of semantic features [95]. Accordingly, semantic features may contribute to an understanding of the contextual text representation in a social media post.…”

Section: E) Lack Of Comparative Evaluationmentioning

confidence: 99%

See 2 more Smart Citations

Twitter Hate Speech Detection: A Systematic Review of Methods, Taxonomy Analysis, Challenges, and Opportunities

2023

View full text Add to dashboard Cite

Hate speech detection has substantially increased interest among researchers in the domain of natural language processing (NLP) and text mining. The number of studies on this topic has been growing dramatically. Thus, the purpose of this analysis is to develop a resource that consists of an outline of the approaches, methods, and techniques employed to address the issue of Twitter hate speech. This study can be used to aid researchers in the development of a more effective model for future studies. This review focused on studies published over the past eight years, i.e., from 2015 to 2022. This systematic search was carried out in December 2020 and updated in July 2022. Ninety-one articles published within the mentioned period met the set criteria and were selected for this review. From the evaluation of these works, it is clear that a perfect solution has yet to be found. To conclude, this paper focused on presenting an in-depth understanding of current perspectives and highlighted research opportunities to boost the quality of hate speech detection systems. In turn, this helps social networking services that seek to detect hate messages generated by users before they are posted, thus reducing the risk of targeted harassment.

show abstract

Section: Discussion and Future Workmentioning

confidence: 99%

Section: E) Separation Of Hate Speech From An Offensive Instancementioning

confidence: 99%

Section: E) Lack Of Comparative Evaluationmentioning

confidence: 99%

See 1 more Smart Citation

Twitter Hate Speech Detection: A Systematic Review of Methods, Taxonomy Analysis, Challenges, and Opportunities

2023

View full text Add to dashboard Cite

show abstract

“…Initial research on hate speech analysis is typically oriented towards monolingual and single classification tasks due to the complexity of the task. They used simple methods such as dictionary lookup [26], bag of words [26], or SVM classifiers [37,57]. Recent efforts are proposing multilingual and multitask learning by using deep learning models [24,43,62,63].…”

Section: Related Workmentioning

confidence: 99%

Understanding and Detecting Hateful Content using Contrastive Learning

González-Pizarro¹,

Zannettou²

2022

Preprint

View full text Add to dashboard Cite

The spread of hate speech and hateful imagery on the Web is a significant problem that needs to be mitigated to improve our Web experience. This work contributes to research efforts to detect and understand hateful content on the Web by undertaking a multimodal analysis of Antisemitism and Islamophobia on 4chan's /pol/ using OpenAI's CLIP. This large pre-trained model uses the Contrastive Learning paradigm. We devise a methodology to identify a set of Antisemitic and Islamophobic hateful textual phrases using Google's Perspective API and manual annotations. Then, we use OpenAI's CLIP to identify images that are highly similar to our Antisemitic/Islamophobic textual phrases. By running our methodology on a dataset that includes 66M posts and 5.8M images shared on 4chan's /pol/ for 18 months, we detect 573,513 posts containing 92K Antisemitic/Islamophobic images and 246K posts that include 420 hateful phrases. Among other things, we find that we can use OpenAI's CLIP model to detect hateful content with an accuracy score of 0.84 (F1 score = 0.58). Also, we find that Antisemitic/Islamophobic imagery is shared in 2x more posts on 4chan's /pol/ compared to Antisemitic/Islamophobic textual phrases, highlighting the need to design more tools for detecting hateful imagery. Finally, we make publicly available a dataset of 420 Antisemitic/Islamophobic phrases and 92K images that can assist researchers in further understanding Antisemitism/Islamophobia and developing more accurate hate speech detection models.

show abstract

“…Senarath and Purohit [25] have ensembled the various diverse features for hate speech detection on two popular Twitter Dataset using SVM. Salminen et al [26] have addressed the lack of model for hate speech detection Inteligencia Artificial 66(2020) across multiple platforms by combining multi-platform data using different feature representations with several machine learning classifiers and neural networks.…”

Section: Related Workmentioning

confidence: 99%

Novel Approach for Generating Hybrid Features Set to Effectively Identify Hate Speech

Shruthi

2020

INT ARTIF

View full text Add to dashboard Cite

Automating hate speech or inappropriate text detection in social media and other internet platforms is gaining a lot of interest and becoming a valuable research topic for both industry and academia in recent years. It is more important for applications to identify the disruptive contents, understand sentiment analysis, identify cyber bullying, detect flames, threats, hatred towards people or particular communities or groups etc. Text classification is a very challenging task due to the nature and complexities with languages, especially its context, micro words, emojis, typo error and sarcasm present in the text. In this paper, we have proposed a model with a novel approach for generating hybrid features for an effective feature representation to classify hate speech. We have combined features learned from deep learning methods with the semantic features like word n-grams and tweets specific syntactic features to form hybrid feature sets. We have also improvised preprocessing steps to reduce the number of missing embeddings to increase the vocabulary for efficient feature learning. We have experimented with the various neural networks for feature learning and machine learning models with hybrid features for classification. Our work delivers hybrid features and appropriate preprocessing techniques for an efficient classification of the standard dataset of 16k annotated hate speech tweets. The combination of Long Short Term Memory (LSTM) trained on Random Embeddings for deep learning features extraction and Logistic Regression (LR) as a classifier with the hybrid features is found to be the best model and it outperforms the state of the art reported in the literature.Resumen La automatización del discurso de odio o la detección de texto inapropiado en las redes sociales y otras plataformas de Internet está ganando mucho interés y se está convirtiendo en un tema de investigación valioso tanto para la industria como para el mundo académico en los últimos años. Es más importante que las aplicaciones identifiquen los contenidos disruptivos, comprendan el análisis de sentimientos, identifiquen el acoso cibernético, detecten llamas, amenazas, odio hacia personas o comunidades o grupos en particular, etc. La clasificación de textos es una tarea muy desafiante debido a la naturaleza y la complejidad de los idiomas , especialmente su contexto, micropalabras, emojis, error tipográfico y sarcasmo presentes en el texto. En este artículo, hemos propuesto un modelo con un enfoque novedoso para generar características híbridas para una representación de características efectiva para clasificar el discurso de odio. Hemos combinado características aprendidas de métodos de aprendizaje profundo con características semánticas como n-gramas de palabras y características sintácticas específicas de tweets para formar conjuntos de características híbridas. También hemos improvisado pasos de preprocesamiento para reducir la cantidad de incrustaciones faltantes para aumentar el vocabulario para un aprendizaje eficiente de funciones. Hemos e...

show abstract

Evaluating Semantic Feature Representations to Efficiently Detect Hate Intent on Social Media

Cited by 18 publications

References 9 publications

Twitter Hate Speech Detection: A Systematic Review of Methods, Taxonomy Analysis, Challenges, and Opportunities

Twitter Hate Speech Detection: A Systematic Review of Methods, Taxonomy Analysis, Challenges, and Opportunities

Understanding and Detecting Hateful Content using Contrastive Learning

Novel Approach for Generating Hybrid Features Set to Effectively Identify Hate Speech

Contact Info

Product

Resources

About