Online offensive behaviour in socialmedia: Detection approaches, comprehensive review and future directions

Chinivar, Sneha; M.S., Roopa; J.S., Arunalatha; K.R., Venugopal

doi:10.1016/j.entcom.2022.100544

Cited by 13 publications

(6 citation statements)

References 73 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Furthermore, advancing research in multimodal fusion techniques and domain-specific feature integration could enhance the robustness and generalization of LSTM-CNN hybrid models across diverse social media platforms and cultural contexts. By effectively integrating textual, visual, and metadata features, hybrid models can capture rich contextual information and linguistic nuances, leading to improved cyberbullying detection performance [44]. Additionally, exploring novel attention mechanisms and adversarial training strategies may further enhance the discriminative power and resilience of LSTM-CNN hybrid models against adversarial attacks and data perturbations [45].…”

Section: Discussionmentioning

confidence: 99%

“…Furthermore, the utilization of domain-specific features has been explored to enhance the effectiveness of hybrid neural network architectures for cyberbullying detection. The research in [21] proposed a feature fusion approach that combines textual, visual, and metadata features extracted from social media posts into an LSTM-CNN hybrid model. This fusion of domain-specific features facilitated a comprehensive analysis of social media content, leading to improved cyberbullying detection performance.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Hybrid Convolutional Recurrent Neural Network for Cyberbullying Detection on Textual Data

Baiganova,

Toxanova,

Yerekesheva

et al. 2024

ijacsa

View full text Add to dashboard Cite

With the burgeoning use of social media platforms, online harassment and cyberbullying have become significant concerns. Traditional mechanisms often falter, necessitating advanced methodologies for efficient detection. This study presents an innovative approach to identifying cyberbullying incidents on social media sites, employing a hybrid neural network architecture that amalgamates Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN). By harnessing the sequential processing capabilities of LSTM to analyze the temporal progression of textual data, and the spatial discernment of CNN to pinpoint bullying keywords and patterns, the model demonstrates substantial improvement in detection accuracy compared to extant methods. A diverse dataset, encompassing multiple social media platforms and linguistic styles, was utilized to train and test the model, ensuring robustness. Results evince that the LSTM-CNN amalgamation can adeptly handle varied sentence structures and contextual nuances, outstripping traditional machine learning classifiers in both specificity and sensitivity. This research underscores the potential of hybrid neural networks in addressing contemporary digital challenges, urging further exploration into blended architectures for nuanced problem-solving in cyber realms.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Hybrid Convolutional Recurrent Neural Network for Cyberbullying Detection on Textual Data

Baiganova,

Toxanova,

Yerekesheva

et al. 2024

ijacsa

View full text Add to dashboard Cite

show abstract

“…In a few cases, it has become so severe that it has led to the extreme of suicide. Therefore, it is important to identify and eliminate offensive behavior as early as possible to make online platforms safer and more secure [30]. For this reason, in recent times, numerous research studies and shares tasks in NLP have emerged regarding the identification of different types of offensive texts.…”

Section: Online Offensive Speechmentioning

confidence: 99%

“…Various online offensive behavior detection techniques in social media can be classified into four main types: content-based, sentiment and emotion-based, user profile-based, and network-based, according to the type of features that the detection techniques address [30].…”

Section: Online Offensive Speechmentioning

confidence: 99%

Comparing Fine-Tuning, Zero and Few-Shot Strategies with Large Language Models in Hate Speech Detection in English

Pan,

García-Díaz,

Valencia-García

2024

CMES

View full text Add to dashboard Cite

Large Language Models (LLMs) are increasingly demonstrating their ability to understand natural language and solve complex tasks, especially through text generation. One of the relevant capabilities is contextual learning, which involves the ability to receive instructions in natural language or task demonstrations to generate expected outputs for test instances without the need for additional training or gradient updates. In recent years, the popularity of social networking has provided a medium through which some users can engage in offensive and harmful online behavior. In this study, we investigate the ability of different LLMs, ranging from zero-shot and few-shot learning to fine-tuning. Our experiments show that LLMs can identify sexist and hateful online texts using zero-shot and few-shot approaches through information retrieval. Furthermore, it is found that the encoder-decoder model called Zephyr achieves the best results with the fine-tuning approach, scoring 86.811% on the Explainable Detection of Online Sexism (EDOS) test-set and 57.453% on the Multilingual Detection of Hate Speech Against Immigrants and Women in Twitter (HatEval) test-set. Finally, it is confirmed that the evaluated models perform well in hate text detection, as they beat the best result in the HatEval task leaderboard. The error analysis shows that contextual learning had difficulty distinguishing between types of hate speech and figurative language. However, the fine-tuned approach tends to produce many false positives. KEYWORDSHate speech detection; zero-shot; few-shot; fine-tuning; natural language processing 1 Introduction Large Language Models (LLMs) are gaining popularity in academic and industrial circles due to their exceptional performance in numerous applications. This success is due to their ability to generate human-like text using huge training datasets and billions of parameters. LLMs have versatile capabilities, unlike previous models that were limited to specific tasks. Due to their remarkable performance in various applications, ranging from general Natural Language Processing (NLP) tasks to specific domain functions, they are increasingly being adopted by various domains, including healthcare, education, law, finance, and scientific research [1].Supervised machine learning consists of training the model on a labeled dataset. However, labeling the dataset is one of the tedious and intensive tasks in NLP. With the emergence of , the

show abstract

“…The following are some prevalent examples of offensive behaviours that people who use social media frequently engage in. Offensive behaviour can take many different forms [9]:…”

Section: A Types Of Offensive Behavior In Social Mediamentioning

confidence: 99%

A Systematic Literature Review on Cyberbullying in Social Media: Taxonomy, Detection Approaches, Datasets, And Future Research Directions

Sahana V

2023

IJRITCC

View full text Add to dashboard Cite

In the area of Natural Language Processing, sentiment analysis, also called opinion mining, aims to extract human thoughts, beliefs, and perceptions from unstructured texts. In the light of social media's rapid growth and the influx of individual comments, reviews and feedback, it has evolved as an attractive, challenging research area. It is one of the most common problems in social media to find toxic textual content. Anonymity and concealment of identity are common on the Internet for people coming from a wide range of diversity of cultures and beliefs. Having freedom of speech, anonymity, and inadequate social media regulations make cyber toxic environment and cyberbullying significant issues, which require a system of automatic detection and prevention. As far as this is concerned, diverse research is taking place based on different approaches and languages, but a comprehensive analysis to examine them from all angles is lacking. This systematic literature review is therefore conducted with the aim of surveying the research and studies done to date on classification of cyberbullying based in textual modality by the research community. It states the definition, , taxonomy, properties, outcome of cyberbullying, roles in cyberbullying along with other forms of bullying and different offensive behavior in social media. This article also shows the latest popular benchmark datasets on cyberbullying, along with their number of classes (Binary/Multiple), reviewing the state-of-the-art methods to detect cyberbullying and abusive content on social media and discuss the factors that drive offenders to indulge in offensive activity, preventive actions to avoid online toxicity, and various cyber laws in different countries. Finally, we identify and discuss the challenges, solutions, additionally future research directions that serve as a reference to overcome cyberbullying in social media.

show abstract

Online offensive behaviour in socialmedia: Detection approaches, comprehensive review and future directions

Cited by 13 publications

References 73 publications

Hybrid Convolutional Recurrent Neural Network for Cyberbullying Detection on Textual Data

Hybrid Convolutional Recurrent Neural Network for Cyberbullying Detection on Textual Data

Comparing Fine-Tuning, Zero and Few-Shot Strategies with Large Language Models in Hate Speech Detection in English

A Systematic Literature Review on Cyberbullying in Social Media: Taxonomy, Detection Approaches, Datasets, And Future Research Directions

Contact Info

Product

Resources

About