A Study of Informatization and Data-Driven Career Planning in Career Guidance in Colleges and Universities

Abstract

Traditional career planning education and employment guidance is based on lectures and text reading, with a large amount of subjective coloring, which can’t achieve one person one strategy. This paper constructs a college career guidance information recommendation system based on collaborative filtering and K-means clustering to recommend suitable career information with a high potential success rate for students from both subjective and objective aspects. The weights of students’ features are objectively calculated, and the K-means algorithm is improved with the minimum spanning tree method to spontaneously calculate the initial number of categories and the class center value from the data, avoiding the problem of random generation and enhancing the effect of clustering. The system is tested against other models in the dataset to examine its clustering performance. It is found that on the optimal K-mean-square error, the errors of the model in this paper are all less than 0.96, which is the lowest among the algorithms on all terms, and its clustering performance is again verified. The best results on the dataset are achieved by the proposed algorithm, with an HR@20 of 59.31%, which is almost twice the SVD recommendation method. The experiment confirmed the significance of each module in the recommender system. Compared with the recommendation of graduates’ jobs, the system is easier to recommend industries for graduates, and the HR indicator is greater than 80% at the recommendation list lengths of 3 and 5, while the HR@5 of SVD, BPR is below 50%. This study provides useful insights into the integration of cutting-edge information technology career guidance and data mining to aid in career planning.

Keywords

• collaborative filtering
• ak-means
• minimum spanning tree method
• career planning
• 94a16