The cloud computing practice has promoted the construction of government cloud with the government‘s cloud technology as the support and integration with multi-sources big data. However, the government cloud is faced with some serious security issues. Many risk factors such as big data leakage, internal abuse, external invasion, and technological loopholes, restrict the effective operation of the government cloud. It is imperative to carry out government cloud security audit based on the risk-oriented model. Through the monitoring of threat behavior and mining security incidents for audit forensics, the cloud model is widely used in the field of e-government. Based on the theoretical review, this paper analyzes the cloud security risk assessment methods under big data and exemplifies the validity of them. It establishes a government cloud security audit framework through the division of responsibilities between cloud providers and cloud tenants, and uses the three dimensions of management, technology, and standards to provide the operational protection of the government's cloud security audit with an aim to find constructive ideas for government cloud security management practices in the era of big data.