Method and system for realizing user-related recommendation

Abstract

The invention provides a method and a system for realizing user-related recommendation. The method comprises: establishing a unified representation of an interaction function for collaborative filtering in a recommendation system; constructing an interactive function search space based on the unified representation of the interactive function; for an input evaluation matrix, determining a corresponding interaction function in the interaction function search space, the row of the evaluation matrix corresponding to a user, the column of the evaluation matrix corresponding to an object, and the matrix element in the evaluation matrix representing an evaluation of the user of the corresponding row of the matrix element on the object of the corresponding column; deriving matrix elements lost in the evaluation matrix based on collaborative filtering using the determined interaction function; and performing a user-related recommendation based on the evaluation matrix in which the lost matrix element has been restored.

Description

Technical field [0001] This application relates to user-related recommendation technology, and more specifically, to a method and system for implementing user-related recommendation using collaborative filtering. Background technique [0002] Recommendation systems are widely used in various scenarios. For example, the recommendation system can use e-commerce websites to provide customers with product information and suggestions, help users decide what products to buy, and simulate sales staff to help customers complete the purchase process. Personalized recommendation is to recommend information and commodities that users are interested in based on the user's interest characteristics and purchase behavior. Recommended objects include commodities, advertisements, news, music, etc. [0003] Collaborative filtering (CF) is a key technology of the recommendation system. Given the evaluation matrix of users and objects, CF aims to predict the missing matrix elements (that is, unknow...

AI Technical Summary

Problems solved by technology

[0008] However, in practical applications, the interaction between the user and the object may be more complex, and the vector inner product as an interaction function may not achieve the best performance for various CF tasks, and operations other than the vector inner product (e.g., vector addition, Vector concatenation, vector outer product, etc., vector convolution) used as an interaction function may sometimes have better performance than vector inner product

[0009] On the other hand, although various vector operations as mentioned above can be used as interaction functions in CF, most of these interaction functions are designed manually, and it is usually not easy to manually select and design appropriate interaction functions for specific CF tasks. and using only a simple operation may not ensure good performance

In addition, with the success of deep networks in various fields, multi-layer perceptrons (MLPs) have also been recently used as interaction functions in CF and achieved good performance, but using MLPs directly leads to difficult architecture selection and is rather time-consuming

Images