The invention provides a B2B platform system. The B2B platform system is characterized by specifically comprising the first step of forming a buyer identity database and a buyer preference database and storing the two databases in a distributed data warehouse, the second step of processing the data obtained from the buyer identity database and the buyer preference database by use of a preference algorithm, and implementing a personalized recommendation method to realizing personalized recommendation under a plurality of scenes, and finally realizing parallel recommendation of different pages of a website to meet the different requirements of users on products at various browsing enquiry stages, and saving the calculation results of the recommendation method in the distributed data warehouse, the third step of providing B2B e-commerce visitors with online efficient data retrieval performance, and the fourth step of displaying the user recommendation results.