Scholarlink

<p>Social media platforms have become part and parcel of our daily communications; however, their commercialization has raised serious concerns regarding user security and privacy. Existing approaches struggle to adapt to evolving cyber threats and often involve high computational costs due to their reliance on large annotated datasets. This study has presented a novel solution, which aims to improve user security and privacy on social media platforms, and an important step in trustworthy content classification and user privacy protection. In order to improve the input representation of our framework, we proposed a new preprocessing method NoTaVe-GloVE. This method provides better semantic representation, which helps with better feature extraction and model performance. The method uses aspect based sentiment analysis to explore users' emotional expressions and predict their behavior. A Bayesian Credibility Model (BCM) is implemented to assess trust scores for the content using various aspects such as user engagement and content quality. The BCM assigns trust scores to contents labeled as most trusted, medium trusted and not trusted, by using a Deep Skill Neural Network (DSNN) which combines Deep Neural Network (DNN) and Skill Optimization Algorithm (SOA), which is established to improve the overall accuracy of classification in social media, based on trust assessment. To ensure immutability, privacy, and secure access, the classified data is stored on a blockchain, ensuring immutability, privacy, and secure access to only authorized entities. This framework aims to address social media security challenges by combining advanced Deep Learning (DL) techniques with blockchain tamper-proof capabilities, providing a comprehensive solution for trustworthy content classification and user privacy preservation. The proposed method achieves 99.5% accuracy and 99.7% precision, demonstrating superior performance compared to existing methods. Its high accuracy and precision highlight its effectiveness in classification tasks. Enhancing adaptability to dynamic threat patterns, reducing computational complexity, and enabling real-time deployment tailored for high-frequency interactions across diverse social media platforms are key future directions.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Secure Data Retrieval and User Behavior Analysis in Social Media Using Blockchain Aware Privacy Preservation

  • 期刊论文

  • 发表日期：

    2025年9月4日

查看全文
  • M. R. Neethu,
  • N. Harini

摘要

Social media platforms have become part and parcel of our daily communications; however, their commercialization has raised serious concerns regarding user security and privacy. Existing approaches struggle to adapt to evolving cyber threats and often involve high computational costs due to their reliance on large annotated datasets. This study has presented a novel solution, which aims to improve user security and privacy on social media platforms, and an important step in trustworthy content classification and user privacy protection. In order to improve the input representation of our framework, we proposed a new preprocessing method NoTaVe-GloVE. This method provides better semantic representation, which helps with better feature extraction and model performance. The method uses aspect based sentiment analysis to explore users' emotional expressions and predict their behavior. A Bayesian Credibility Model (BCM) is implemented to assess trust scores for the content using various aspects such as user engagement and content quality. The BCM assigns trust scores to contents labeled as most trusted, medium trusted and not trusted, by using a Deep Skill Neural Network (DSNN) which combines Deep Neural Network (DNN) and Skill Optimization Algorithm (SOA), which is established to improve the overall accuracy of classification in social media, based on trust assessment. To ensure immutability, privacy, and secure access, the classified data is stored on a blockchain, ensuring immutability, privacy, and secure access to only authorized entities. This framework aims to address social media security challenges by combining advanced Deep Learning (DL) techniques with blockchain tamper-proof capabilities, providing a comprehensive solution for trustworthy content classification and user privacy preservation. The proposed method achieves 99.5% accuracy and 99.7% precision, demonstrating superior performance compared to existing methods. Its high accuracy and precision highlight its effectiveness in classification tasks. Enhancing adaptability to dynamic threat patterns, reducing computational complexity, and enabling real-time deployment tailored for high-frequency interactions across diverse social media platforms are key future directions.