User dynamic preference oriented commodity sequence personalized recommendation method

Abstract

The invention discloses a user dynamic preference oriented commodity sequence personalized recommendation method. The method comprises the steps of extracting a commodity sequence under the same usersimilarity score to construct a commodity score vector; extracting a user sequence of similar scores of the same commodity to obtain a user score vector; combining user personal information, commoditybasic attribute information, user and article comments and commodity pictures; achieving feature extraction of a user and a commodity based on multi-task learning, taking a user feature vector and afeature vector of a historical commodity sequence of the user feature vector as input, achieving generation of the commodity sequence by training a coder-decoder, and accurately learning recommendation of the optimal commodity sequence in combination with a search strategy. According to the invention, based on the multi-modal user-commodity data, the user features and the commodity features are highly extracted and fused, personalized recommendation of the commodity sequence for user preferences is realized, and the user experience is improved.

Description

technical field [0001] The invention relates to the field of intelligent commerce, in particular to a method for personalized recommendation of product sequences oriented to user dynamic preferences. Background technique [0002] With the advent of the era of big data, a large number of users generate a large amount of information all the time, and these information items are very large. How to effectively extract effective features from the huge data for recommendation is a problem that needs to be solved urgently. In the field of recommendation, the information preferences of different users are different, but there is a certain correlation between similar users and products. How to propose an efficient and reasonable personalized recommendation method in the context of big data is a hotspot worthy of research. [0003] Existing recommendation algorithms often use the relationship matrix between users and products and combine them with other features to make recommendatio...

AI Technical Summary

Problems solved by technology

[0003] Existing recommendation algorithms often use the relationship matrix between users and products and combine them with other features to make recommendations. Although they can provide some reference for the recommendation results, these recommendation systems have the following disadvantages: they are not suitable for big data The processing efficiency of a large amount of user and item information is low, and it is easy to ignore the dynamic historical preferences of users and the evaluation of historical products by users, and cannot provide good feedback on previous recommendation results; secondly, the traditional recommendation system model is simple and user characteristics Single, the use of probability and statistics methods for evaluation has certain limitations, the recommendation time is long, and the error is large; the traditional recommendation system does not combine the picture features of the product, the task is single, and it is impossible to qualitatively evaluate the user's historical comments and product pictures. Therefore, it is impossible to effectively evaluate the recommended features; finally, the traditional recommendation algorithm ignores the sequence information between the product sequence with the same user rating and the user sequence corresponding to the product, and the recommendation results are not properly sorted and fused. , resulting in a poor overall recommendation effect

Images