The increasingly severe employment situation requires higher vocational colleges to optimize the means of talent cultivation. This paper collects enterprise recruitment data through web crawler technology as the research data base of cross-border compound music talent skill cultivation. Using the word frequency-inverse document frequency (TF-IDF) and Word2Vec word embedding model, the recruitment data are processed to retain the job keywords. Mining association rules between job recruitment data and skill requirements. According to the results of association rule mining, the optimization scheme of cross-border composite music talent cultivation in higher vocational colleges is proposed. The study shows that the 6 job categories that provide the most positions for crossborder composite talents require talents to master 2 or more skills. The 6 strong association rules for each of the 3 types of cities show that the probability of different recruiting positions requiring composite talents to master a certain type of skill ranges from 80% to 100%, which is a mandatory requirement and requires more attention from institutions and job seekers.
