Abstract: A blockchain sharding method combining spectral clustering and a reputation value mechanism includes: obtaining, every other account grouping cycle Ta, account transaction data recorded during operation of a blockchain to generate an account transaction graph; obtaining an adjacency similarity matrix W and a degree matrix D based on the account transaction graph; generating a Normalized Laplace matrix L based on the adjacency similarity matrix W and the degree matrix D, performing dimension reduction on L to obtain a feature matrix F, and then clustering the feature matrix F with a clustering dimension of k by row through a K-means clustering method; and dividing blockchain accounts into k groups based on an obtained clustering result, and allocating accounts in the k groups to k blockchain shards.