Research on Blockchain Data Sharing and Privacy Protection Scheme Based on Reputation Value Calculation

The issues of data sharing and privacy protection have always been a major research focus in blockchain. In response to problems such as data security and personal privacy protection that are prevalent in medical big data, in order to protect patients' privacy while efficiently utilizing the medical data of various medical institutions, a blockchain data sharing and privacy protection scheme based on reputation value calculation is proposed. By screening members with high - quality data before training, the data sharing process is transformed into a model parameter sharing process, avoiding the upload of original data to the blockchain and thus preventing patient privacy leakage. At the same time, the reputation scheme is improved to select candidate hospitals with high-quality data and reliability to participate in the federated learning training model. The distributed consortium blockchain is used to store the reputation values of each hospital and the global model parameters, and hospitals that actively contribute high-quality data are rewarded. The experimental results show that the proposed scheme can screen out high-quality data sources, reduce the interaction time between edge nodes and malicious nodes. Even when the reputation value is above 0.5, a learning accuracy of 0.857 can be achieved. Compared with existing typical federated learning schemes, the performance of the proposed scheme is superior.