With the development of globalization, the cross-cultural market is facing needs such as diversification and personalization of consumer demand. Based on the theory of market segmentation, the study proposes an ant colony algorithm to improve the market segmentation model of K-means clustering, and examines its effectiveness. Further, a personalized recommendation algorithm based on multivariate dynamic user profiles is proposed to recommend products to target users more accurately. A reliable simulation environment is constructed based on the KuaiRec dataset and the classical LastFM dataset to properly evaluate the performance and effectiveness of the model on the recommendation platform. Through the K-means ant colony clustering algorithm proposed in this paper to divide the interest information and attribute information of users, the users as a whole are classified into specific categories, and the online_reward value of the personalized recommendation algorithm based on multivariate dynamic user profiles proposed in this paper fluctuates from 50.05 to 50.49, which is a significantly superior performance. As a result, this paper concludes that crosscultural marketing strategies should be marketed at four levels: product, price, channel, and promotion, in order to adapt to regional cultures, attract consumers, and build consumer loyalty and satisfaction.