Hybrid Approach for the OSN Privacy Preservation using Artificial Intelligence
Sarika Vasantrao Bodake, Pooja Sharma, Sandeep Kadam · 2024
Real-time streaming data mining is harder than static data mining because it processes massive, unstructured streams of data. The inclusion of sensitive data in streaming data perpetuates the privacy issue. Static data anonymization research has advanced greatly in recent years. Suppression and generalisation remove quasi-identifiers from data. Since streaming data is dynamic and may have infinite properties, it is difficult. Despite various network and user privacy preservation methods, OSN researchers still struggle to achieve the k-anonymity, l-diversity privacy criterion. We anonymize OSNs in this study via clustering based on many network properties. Clustering aims to implement privacy for OSN network edges, nodes, and user characteristics. The proposed model's clustering strategy ensures k-anonymity and l-diversity in each cluster. We start by creating the data normalisation method to pre-process and enhance OSN data. The next stage in k-anonymization is clustering OSN data by graph characteristics. Additionally, a novel one-pass anonymization method improved the k-anonymization clusters, which satisfied l-diversity privacy norms. We evaluate the recommended technique using cutting-edge methods on a real-world dataset. In terms of anonymization, information loss, and execution time, the recommended method provides high-level privacy compared to current methods.