Serialized recommendation method based on long-term and short-term interests

Abstract

The invention provides an educational background deterioration recommendation method based on long-term and short-term interests, and the method comprises the steps: processing user purchase sequencedata and user questioning data in a data set, obtaining the serialized interaction data of a user and a commodity, and extracting the comment content of the user for the commodity to represent the features of the commodity; learning stable long-term preferences of the user from historical purchase sequence data of the user by using a recurrent neural network, and modeling instant interests of theuser by using questioning data; and finally, for stable long-term preferences and dynamic instant interests, using an Attention mechanism to describe the degree of dependence of different users on thetwo features. The problem of inaccurate recommendation caused by user preference evolution can be effectively solved, and meanwhile, the dependence degrees of different users on long-term preferencesand different instant interests can be effectively expressed.

Description

technical field [0001] The present invention relates to the field of serialized recommendation and recommendation system based on deep learning, in particular to a method for serialized commodity recommendation based on long-term and short-term preferences. Background technique [0002] As an important part of modern e-commerce websites, recent recommendation systems try to recommend items that users want to buy or interact with in the future based on their interests or preferences. With the development of e-commerce mechanism, a large number of user interactions (such as browsing, clicking, collecting, shopping cart, purchasing) are recorded, which hides the consumption patterns of users. These logs with sufficient information provide a data basis for researching user preferences and personalized recommendations. [0003] Existing recommender systems can be summarized in two main ways to model the interaction between users and items. The first approach is matrix factoriza...

AI Technical Summary

Problems solved by technology

[0005] Although the existing sequence models can predict the items that the user may purchase in the next purchase based on the interaction behavior sequence, there are two shortcomings: first, these methods focus on directly using the sequence between items to represent the relationship between items, However, since different users pay attention to different aspects when purchasing the same product, the item vector directly represented by the similarity between items cannot directly represent the user's preference. Second, the existing models ignore the user's immediate interest. Unlike short-term preferences of users, immediate interests are immediate, specific needs when users want to buy a product or a series of products

Images