As an important foundation of materials genome engineering, the establishment of materials database and the optimization of process parameter have an important impact on materials research and development. To address current challenges such as traditional relational databases being inadequate for storing multi-source, high-dimensional, and heterogeneous alloy data, and the urgent need to shift from the traditional trial-and-error experimentation to a data-driven research paradigm, this paper proposes to establish an alloy graph database based on ontology and graph data models. Based on this foundation, an improving dung beetle optimization (IDBO) algorithm is introduced to optimize combinations of process parameters. The algorithm enhances global search capability by initializing the population using inverse learning and updating dung beetle position using a variable spiral search strategy. To avoid local optima, Cauchy mutation perturbation and hybrid strategies are fused around the optimal solution to generate new candidates. To verify the performance of the proposed IDBO algorithm, it was tested on 23 classical benchmark functions, with results showing that IDBO algorithm outperforms other algorithms and significantly improves the convergence speed and optimization accuracy. Furthermore, when applied to the optimization of alloy process parameters, the IDBO algorithm not only demonstrated superior performance compared to other optimization algorithms but also successfully identified the optimal parameter combinations, validating its superiority in process parameter optimization.