Abstract: In recent years, the lines between social networking and e-commerce have steadily blurred. Many e-commerce websites let consumers sign in using their Facebook or Twitter accounts or other social networking profiles. New purchases may also be announced on the microblog, along with links to the product pages on e-commerce websites. In this study, we suggest a novel method to the problem of promoting products from e-commerce sites to users on social media in "cold start" conditions. When it comes to making suggestions for new products on social networking sites, the data is quite difficult to work with. Users linked via social networking sites and e-commerce websites may be used to convert social networking characteristics to refer to another component of the product offer (users who have social networking accounts and make purchases on e-commerce sites). Furthermore, we propose training consumers and embedding products from e-commerce site data using repeated neural networks (also known as "embedding", "coding," or "embedding"), and translating users' social networking using gradient-enhanced trees. components that go into consumer weddings. We create feature-based matrix factor technology for Cold Start product recommendations based on user marriages. The enormous Chinese microblogging service SINA WEIBO and the sizable dataset produced by the biggest Chinese B2C e-commerce site JINGDONG serve as examples of how the suggested framework is used.

| DOI: 10.17148/IARJSET.2022.9720