In the context of the digital era, the real-time and accuracy of enterprise accounting costing has become the key to improve the efficiency of decision-making. This paper proposes a comprehensive method system integrating rough set theory and attribute approximation algorithm, aiming at reconstructing the framework of enterprise cost accounting through technological innovation and process optimization. Aiming at the problems of low efficiency and redundant accounts in the existing computerized accounting system, the paper proposes the development of customized accounting software and the strategy of setting up cost items scientifically to optimize the classification and aggregation of direct materials, labor and auxiliary expenses. It also introduces the attribute simplification algorithm of rough set theory, reduces data redundancy through knowledge granularization and dimensionality reduction technology, and combines with the heuristic greedy algorithm to realize dynamic cost tracking and refined management. The empirical part takes Building Construction Enterprise A as a case study, and the relative error of the data mining-based procurement cost accounting system is reduced to 0.175%, which is significantly better than the traditional method (1.135%). Meanwhile, the attribute approximation algorithm is outstanding in memory performance, compared with Genmax, MAFIA and other algorithms, the memory consumption is reduced to 263.59 in a huge scale dataset (700,000 records) with 20% support, which verifies its high efficiency and applicability in complex data scenarios.
