Intelligent Material Code Deduplication and Governance Using Tokenization-Based Fuzzy Matching for Industrial Group Collaboration

Ensuring the uniqueness of material codes in group-level material management is essential for integrating business and finance across an industrial supply chain. This study proposes an intelligent deduplication and governance approach by incorporating a fuzzy matching algorithm based on tokenization matrices. The method enhances material management efficiency by computing the similarity between newly applied material codes and historical records, ensuring precise duplication detection. The algorithm tokenizes material names, specifications, and models to construct a tokenization matrix, enabling accurate similarity calculations. It is applied to key approval processes in the aluminum industry, including material code application, deactivation, activation, and merging. Empirical results from a large aluminum industry group demonstrate that the algorithm significantly reduces duplicate codes, improving data accuracy and management efficiency. The findings show that this approach optimizes inventory balancing, minimizes overstocking, and reduces capital occupation, contributing to cost control and operational efficiency. Moreover, it supports cross-subsidiary material sharing, reinforcing data standardization across departments and systems. The proposed method offers a scalable and effective solution for intelligent material governance in industrial groups, highlighting its broad applicability and high promotional value in enterprise-wide material management.