Research on Knowledge Management Optimization of Enterprise Digital Transformation Based on Multi-dimensional Feature Mining of Domain Ontology

There are various complex interactive relationships between the knowledge of enterprise digital transformation, which leads to the inability of ontology to fully reflect the true structure of enterprise knowledge, thereby affecting the integration and utilization of knowledge. Therefore, a knowledge management optimization method for enterprise digital transformation based on multidimensional feature mining of domain ontology is proposed. First, an ontology tailored for knowledge management within the realm of enterprise digital transformation was developed, encompassing classes, relationships, attributes, and their respective instances. This was meticulously crafted using the Protégé tool. Second, by employing techniques such as phase space reconstruction, wavelet decomposition, and empirical mode decomposition (EMD), we can delve deeply into the nonlinear time series and multi-scale characteristics of knowledge data pertaining to enterprise digital transformation. By constructing features, the extracted multidimensional features are transformed into a set of feature vectors, providing strong support for optimizing knowledge management in enterprise digital transformation. Finally, a compact radial basis function is introduced to construct a multidimensional data reconstruction framework, calculate the contribution coefficients and weights of sampling points, and lay the foundation for optimizing knowledge management in enterprise digital transformation. Based on the weights of data sampling points and estimated values of influencing factors, management optimization parameters are introduced to achieve optimized management of knowledge data for enterprise digital transformation. The test results reveal that the method introduced in this paper achieves an impressive average feature interaction gain of 4.93 in enhancing knowledge management for enterprise digital transformation. With a knowledge utilization rate peaking at 97% and a consistent knowledge coverage rate exceeding 82%, the proposed method exhibits outstanding performance.