Enhanced cross-domain sequential recommendation method combining large language model and comparative learning

The invention discloses an enhanced cross-domain sequence recommendation method combining a large language model and comparative learning, which belongs to the field of cross-domain sequence recommendation, and comprises the steps of fusing the large language model, finely adjusting a fusion proportion, decoupling long and short features of a sequence, respectively fusing long-term features and short-term features of different domains, and obtaining long-term interests and short-term interests. And fusing the long-term interests and the short-term interests of different domains after fusion to predict a next item. According to the method, effective fusion of long-term interests and short-term interests is realized, so that the accuracy of cross-domain sequential recommendation is remarkably improved. The accuracy of cross-domain sequence recommendation is improved to a certain extent, and a new view angle and a new technical path are provided for research and practice of a recommendation system. 本发明公开了一种结合大语言