Learning Wisdom from Crowds: A Trustworthy and Robust Multi-label Annotation Aggregation Platform atop Blockchain - RMGS

Multiple-label annotation has been extensively applied in various fields, which could be highly related to our daily life, e.g., information retrieval, goods recommendation, medical diagnose. Blockchain, with the character of transparency and decentralization, emerges to an ideal place for multi-label aggregation. To continuously provide high-quality labels, we aim to leverage the wisdom of crowds to build a secure and robust multi-label annotation aggregation platform atop the blockchain.However, untrustworthy issues become more prominent in the decentralized setting. Bad behaviour of clients, low-quality labels provided by clients can drastically degrade the usability of the multi-label annotation aggregation platform. In this project, we propose to investigate the accompanying security issues and corresponding countermeasures. Our methodology will leverage existing quality-control approaches and security constructs, algorithmic, computational techniques to achieve our goal. The study would promote the label-enhanced application and provide insights for research on secure aggregation for various data atop the blockchain.