A Multilayered Preprocessing Approach for Recognition and Classification of Malicious Social Network Messages

Čepulionytė, Aušra; Toldinas, Jevgenijus; Lozinskis, Borisas

doi:10.3390/electronics12183785

Cited by 3 publications

(1 citation statement)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Various methods, including Logistic Regression, K-Nearest Neighbor, SVM, and Decision Tree, have been adapted for multi-label classification problems [41,42]. These methods transform the multilabel problem into a binary classification task, achieving high accuracy and f1-scores, although some bias towards non-toxic classes has been observed [43,44].…”

Section: Literature Reviewmentioning

confidence: 99%

Graph convolution networks for social media trolls detection use deep feature extraction

Asif,

Al-Razgan,

Ali

et al. 2024

J Cloud Comp

View full text Add to dashboard Cite

This study presents a novel approach to identifying trolls and toxic content on social media using deep learning. We developed a machine-learning model capable of detecting toxic images through their embedded text content. Our approach leverages GloVe word embeddings to enhance the model's predictive accuracy. We also utilized Graph Convolutional Networks (GCNs) to effectively analyze the intricate relationships inherent in social media data. The practical implications of our work are significant, despite some limitations in the model's performance. While the model accurately identifies toxic content more than half of the time, it struggles with precision, correctly identifying positive instances less than 50% of the time. Additionally, its ability to detect all positive cases (recall) is limited, capturing only 40% of them. The F1-score, which is a measure of the model's balance between precision and recall, stands at around 0.4, indicating a need for further refinement to enhance its effectiveness. This research offers a promising step towards more effective monitoring and moderation of toxic content on social platforms.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

Graph convolution networks for social media trolls detection use deep feature extraction

Asif,

Al-Razgan,

Ali

et al. 2024

J Cloud Comp

View full text Add to dashboard Cite

show abstract

CBDC-Net: Recurrent Bidirectional LSTM Neural Networks Based Cyberbullying Detection with Synonym-Level N-Gram and TSR-SCSOFeatures

P. Padma,

G. Siva Nageswara Rao

2024

IJCESEN

View full text Add to dashboard Cite

Social networks Cyber bullying has become another common problem in online social networks (OSNs) which exposes individuals to high risks of their mental health and interacting with others. Previous work in cyber bullying detection is often confronted with limitations in accurately detecting abusive behavior because of the intricacies in cyber space and evolution of cyber bullying practices. A new approach of Cyber bullying detection and classification network (CBDC- Net) for improving the effectiveness of detection of cyber bullying in OSNs based on natural language processing features, feature selection techniques, and deep learning algorithms is also presented in this study. CBDC-Net can overcome these challenges to existing detection methods of cyber bullying using innovative Natural Language Processing (NLP) and Deep Learning approaches. In the data preprocessing step, CBDC-Net filter and normalize the text data that is openly collected from OSNs. After that, CBDC-Net extracts features using a Synonym Level N-Gram (SLNG) approach and it incorporates both the word and character-based information to make the synonyms of text much better than the other method. After that, CSI of CBDC-Net applied Textual Similarity Resilient Sand Cat Swarm Optimization (TSR-SCSO) for feature selection to give an iterative value of their features’ importance level to detect cyber bullying. Last, in CBDC-Net, a Recurrent Bidirectional Long Short-Term Memory (LSTM)Neural Network for classification (RBLNN) is used as classification approach is applied, which recognizes the sequential nature of textual data enabling proper distinction between cyber bullying cases. Last but not the least, the CBDC Net provides a promising solution for solving the mentioned problems of cyber bullying detection in OSNs.

show abstract

Safeguarding Online Communications using DistilRoBERTa for Detection of Terrorism and Offensive Chats

Shah,

Abuaieta,

Almazrouei

2024

JISCR

View full text Add to dashboard Cite

People use social media for both good and distasteful purposes. When used with malicious intent, it raises significant concerns as it involves the use of offensive language and hate speech that promote terrorism and other negative behaviors. To create a safe, secure and pleasant environment, these communications must be closely monitored to prevent severe problems, associated risks and other pertinent issues. With the help of AI, specifically Large Language Models (LLM), we can quickly analyze text and speech to determine whether the communications promote the dangers identified here above not to mention other toxic elements. For this research, the LLM used is the DistilRoBERTa model from the Transformers library using Hugging Face. The DistilRoBERTa model was trained on datasets consisting of terrorism-related conversations, offensive-related conversations, and neutral conversations. These datasets were obtained from publicly available sources. The results of the experimentation show that the model achieved 99% accuracy, precision, recall, F1 score, and ROC curve. To improve the robustness of the model, it must be continuously fine-tuned to predict dynamic communication behavior since real conversations are inaccessible due to restrictions. A drag-and-drop interface is used to upload the files and get the categorical output, ensuring seamless and easy interaction.

show abstract

A Multilayered Preprocessing Approach for Recognition and Classification of Malicious Social Network Messages

Cited by 3 publications

References 45 publications

Graph convolution networks for social media trolls detection use deep feature extraction

Graph convolution networks for social media trolls detection use deep feature extraction

CBDC-Net: Recurrent Bidirectional LSTM Neural Networks Based Cyberbullying Detection with Synonym-Level N-Gram and TSR-SCSOFeatures

Safeguarding Online Communications using DistilRoBERTa for Detection of Terrorism and Offensive Chats

Contact Info

Product

Resources

About