2024/02/03 by Tanveer Khan, Fahad Sohrab, Khan, Tanveer +5 · 1 citation
Computer Science · Social Sciences · #Access Control and Trust #Artificial Intelligence (cs.AI) #Ethics and Social Impacts of AI #FOS: Computer and information sciences #Privacy-Preserving Technologies in Data #Social and Information Networks (cs.SI)
paper · pdf · doi:10.48550/arxiv.2402.02066
openalex publication_date 2024/02/03 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
\mathbbX (formerly Twitter) is a prominent online social media platform that plays an important role in sharing information making the content generated on this platform a valuable source of information. Ensuring trust on \mathbbX is essential to determine the user credibility and prevents issues across various domains. While assigning credibility to \mathbbX users and classifying them as trusted or untrusted is commonly carried out using traditional machine learning models, there is limited exploration about the use of One-Class Classification (OCC) models for this purpose. In this study, we use various OCC models for \mathbbX user classification. Additionally, we propose using a subspace-learning-based approach that simultaneously optimizes both the subspace and data description for OCC. We also introduce a novel regularization term for Subspace Support Vector Data Description (SSVDD), expressing data concentration in a lower-dimensional subspace that captures diverse graph structures. Experimental results show superior performance of the introduced regularization term for SSVDD compared to baseline models and state-of-the-art techniques for \mathbbX user classification.