Order recommendation method and device

The invention provides an order recommendation method, and the method comprises the steps: obtaining a plurality of historical orders, carrying out the Hash coding of the text information of each historical order, and obtaining a historical related order set of each historical order; according to the order information of each historical order and the order information of the N historical related orders with large similarity of the historical order, obtaining a similarity feature matrix of the historical order, and obtaining a similarity feature matrix training set to obtain an order sorting model; and inputting the similarity feature matrix of the current order into an order sorting model to obtain N recommended orders related to the current order. According to the invention, through mining user data and behaviors in practical application, enterprises are helped to sell commodities, economic benefits are increased, and time and effort for users to select commodities are saved. 本发明提供一种订单推荐方法，该方法包括：获取多个历史订单，对每个