One Person One Vote: Achieving Temporal Dynamic and Byzantine-Resilient Digital Community

Abstract

Digital communities are dynamically developed with users admitted in as digital identities, and process their affairs via egalitarian decision processes, namely one person one vote. However, the digital democracy in these digital communities is threatened by Byzantines therein. Most existing works focused on Byzantine detection, but we are interested in growing Byzantine-resilient community rather than whitelisting. Several works concerning developing a Byzantine-resilient digital community are vulnerable to the collapse of these selected digital identities or impractical binarized trust relations among digital identities. To this end, we propose two practical schemes based on edge links and attributes that can achieve temporal dynamic and Byzantine-resilient digital communities, providing digital democracy. Specifically, we first propose the mixed sampling of links and attributes in digital community to output node-edge sequences. Then, we further design the skip gram-based quantification of trust relationships using the node-edge sequences. Thereafter, based on the quantified trust relationships, we propose vertex-based and edge-based strategies that prove the constraints when dynamically developing a Byzantine-resilient digital community. The key advantage is that our work can be applied to any graph containing both digital identity nodes and attribute nodes, rather than the graphs with one kind node and the fully connected graphs. Last, we conduct experiments on four real-world datasets, and the extensive results indicate the superior performance of our work, compared to four existing works. This work can be applied to social networks, online shopping platforms, etc., and keep digital democracy therein.