This paper utilizes data mining methods to mine talent profiles. Through the Scrapy framework and text mining methods, talent tags are mined and their features are extracted. To address the limitations of the traditional FCM clustering algorithm, the CFSFDP algorithm is introduced for optimization, proposing a density peak-optimized fuzzy C-means algorithm (FDP-FCM). This algorithm is compared with other algorithms in terms of clustering performance and robustness to evaluate its effectiveness. Regression analysis is employed to explore the influencing factors and differences in graduates’ job-seeking and further education decisions. Finally, a talent cultivation model for traditional cultural integration and innovation based on user profiles is proposed. The FDPFCM algorithm achieved the best performance among all algorithms in terms of F-measure, RI, and Jaccard coefficient. Under the new productive forces, provincial types have predictive effects on the achievement of employment and further education goals for graduates. Academic performance and job preparation have more significant predictive effects on employment-oriented graduates, while academic performance has a more pronounced predictive effect on further education-oriented graduates.
