A Recommendation Method Based on Deep Social Relationships

Abstract

The invention discloses a recommendation method based on in-depth social relationships, including: 1. Constructing a user's rating matrix for items and a social relationship matrix between users; 2. Constructing an input layer through one-hot encoding; 3. According to The social relationship matrix between users is automatically encoded to obtain the social embedding matrix; 4. Construct the embedding layer through the input layer and the social embedding matrix; 5. Construct the synergy layer through the embedding layer; 6. Input the synergy layer information into the hidden layer , to get the corresponding prediction score information, so as to recommend items. The invention can make full use of the social relationship between users to solve the problem of data sparsity, and at the same time, establish a complex deep neural network structure between users and items to ensure good accuracy, and then accurately recommend users.

Description

technical field [0001] The invention relates to the field of personalized recommendation, in particular to a recommendation method based on deep social relationship. [0002] technical background [0003] With the advent of the era of information explosion, recommendation systems are used in various industries, including: movies, books, academic papers, etc. The recommendation system mainly recommends items for each user independently based on the user's past purchase records and search history, as well as the behavior of other users. [0004] Collaborative filtering algorithms are widely used in the field of recommendation systems. The collaborative filtering algorithm discovers the user's preferences by mining the user's historical behavior data, divides the users into groups based on different preferences, and recommends items with similar tastes. Although the collaborative filtering algorithm is widely used and has high precision, it has the problem of data sparsity. ...

AI Technical Summary

Problems solved by technology

[0006] In order to solve the problem of data sparsity in existing collaborative filtering algorithms, the present invention proposes a recommendation method based on deep social relationships, hoping to make full use of the social relationships between users to solve the problem of data sparsity, and at the same time establish a complex relationship between users and items The deep neural network structure ensures good accuracy, and then realizes accurate recommendations for users

Images