Application Research of Dataset Analysis and Optimization Method Based on Artificial Intelligence Technology in University Talent Training System

To address the serious homogenization of talent cultivation in universities and the insufficient social adaptability and contribution potential of students, a personalized university talent cultivation model driven by artificial intelligence has been proposed. The study aims to optimize data collection and processing using artificial intelligence technology by calculating itemset boundary values and keyword thresholds, and then construct a multidimensional evaluation index system that includes talent cultivation quality, educational resource utilization, and social adaptation and contribution. By utilizing techniques such as keyword extraction, similarity calculation, and convolutional neural network classification, the personalized talent training model has been evaluated and optimized, achieving precise cultivation of students’ comprehensive qualities and innovative abilities. This provides scientific basis and technical support for the reform and development of talent training models in universities. The experimental results show that in the courses evaluated by the method proposed in this paper, the scores of higher mathematics in terms of talent cultivation quality, educational resource utilization, and social adaptation and contribution are 0.95, 0.92, and 0.94, respectively, indicating the best performance, and other courses also score above 0.8. The evaluation result of the construction effect of personalized university talent training mode is close to 1, showing an ideal construction effect. At the same time, 50 rules were mined under the optimal threshold setting for association rule mining. The CNN model achieved high efficiency of processing 230 samples per second, and a high classification accuracy of 92% when processing learning behavior data.