Application of Data Security Models Based on Neural Network Algorithms in Optimizing Social Network Order

Internet data is changing government management, business operations, public services, and people's lives, and driving social change. From the perspective of social management, the Internet data era has brought opportunities for innovation and development, but also serious challenges. This paper aimed to study how to use neural network algorithms. This paper analyzed and designed a data security evaluation model for the Internet, and improved the Back Propagation (BP) algorithm. This article compared and analyzed the BP algorithm and the improved BP algorithm in experiments, starting with the accuracy and time consumption of network data security evaluation. The data shows that when the number of test samples was 400, the accuracy rates of the two algorithms were 90.89% and 99.18%, respectively, and the consumption time was 3.21s and 1.56s, respectively. It is not difficult to see that the improved BP algorithm plays a more excellent performance in the network data security evaluation model.