Optimization of data model-driven design thinking in the software development process

Model-driven software development has become a hot research topic and discovery trend in the field of software engineering. Its core idea is to treat analysis and design models as equivalent to code. Better integration of models and code can greatly increase the chances of effective improvement and achieve automated software development through abstract models. In this paper, we first constructed a data model-driven architecture system based on the meta-modeling hierarchy, using a data dictionary for data storage. The mapping relation loader transforms the data extracted from the dictionary. Using the differential evolution algorithm, the model is defined as a metamodel that actually exists. At the same time, the MapReduce framework is combined to parallelize the computation of the DE algorithm based on the island model in order to solve the problem of poor optimization of the differential algorithm. Apply the model to actual software development and realize data visualization and display using Flex technology. Simulation experiments are set up to test the performance of the model and the platform. After CSI uncertainty estimation, the MSE index of the model is analyzed. This paper’s data model-driven method, with an MSE value of only 0.01084, stands out among the five methods in Case 4. Under the condition of 300 users concurrently, the user’s access response time is tested, and it can be seen that the user’s response time is within 1~2.9s, which passes the performance test. © 2024 Yun Li and Lianwei Li, published by Sciendo.