Digital transformation has a profound impact on higher education, and the training of business administration professionals faces new challenges. The demand of enterprises for business administration talents has undergone structural changes, requiring digital literacy and innovation ability. This study constructs a curriculum system for business administration majors based on learning path mining algorithm under digital talent cultivation mode. Methodologically, the traditional DINA model is improved, reaction time parameters are introduced, and a reparameterized DINA model is established; a core literacy assessment framework containing cognitive model construction, Q matrix establishment, data collection, and diagnostic output is designed; potential knowledge states are identified through clustering analysis, and learning paths are portrayed. The results show that for the test of 379 business administration students, the average score is 12.665, the probability of mastering organization management (A1) reaches 0.9, and human resource management (A4) is only 0.4; clustering analysis yields 8 categories of knowledge states, with KS3 accounting for 30.08% as the most; 4 complete learning paths are identified, with path 2 covering 226 people accounting for 59.6%. The conclusion shows that the improved cognitive diagnostic model can effectively identify the differences in students’ knowledge mastery, and the learning path mining provides the basis for personalized teaching; accordingly, it proposes the constructive strategies of reconstructing the curriculum system, preparing the syllabus, and realizing the theoretical-practical docking to push forward the reform of digital talent cultivation for business administration majors.
