Natural Disaster Resilience Approach (NDRA) to Online Social Networks

Abstract

Numerous disasters impact people around the world, including floods, cyclones, tsunamis and so on. Disaster resilience (recovery) is a realtime challenge. The research goal of the proposed Natural Disaster Resilience Approach (NDRA) is to fasten and automate disaster resilience within the affected region. Prioritization for saving lives must be provided to the region most affected. Currently almost all people use the Online Social Networks (OSNs) to share their daily activities. Actually only their friends in OSN can note the photos posted by OSN users in the disaster-affected region and in turn can support the friends in need. Our research goal is to conduct speedier and prioritized disaster resilience processes by directly addressing honest disaster help requests from images posted by OSN users. To achieve the above goal, we suggest an approach known as the Natural Disaster Resilience Approach (NDRA). The NDRA is a three-tiered framework: Tier-1: sybil(malicious) user avoidance engine: prevents fake OSN user profile creation; Tier-2: ‘D’- attributed image classifier: determines the originality of the images posted and Tier-3: sybil user prediction: uses an Advanced Sybil Node Prediction Algorithm (ASYNPA) to check the authenticity of the user and priorities for faster resilience requests. In NDRA we use Advogato dataset with 6541 users and 51,127 edges. Ultimately, the comparison is only made between the proposed ASYNPA tier-3 and the existing VoteTrust algorithm, and the graph is plotted against the False Positive (FP) rate, taking into account the precision and recall metrics. Around 99.84% of the expected sybils were confirmed in ASYNPA, which is 3.49% higher than VoteTrust.