Personalized product recommendation method based on combined non-negative matrix decomposition
A technology of non-negative matrix decomposition and recommendation method, which is applied in marketing and other directions, and can solve problems such as inability to effectively deal with new users
Patent Information
- Authority / Receiving Office
- CN · China
- Current Assignee / Owner
- Publication Date
- 2014-02-05
- Estimated Expiration
- Not applicable · inactive patent
Smart Images
Figure 1 Figure 2 Figure 3
Abstract
Description
technical field
[0001] The invention relates to the technical fields of non-negative matrix factorization and product recommendation, especially considering the complex social network structure of users and the product recommendation work of joint non-negative matrix factorization. Background technique
[0002] With the rapid development of the Internet, more and more physical goods are turning to online sales. Online sales saves the investment in stores for physical sales, reduces the labor cost of store maintenance, and at the same time it is easier to get rid of geographical restrictions and sell products to all parts of the country and even other countries. However, in the face of a large number of potential users, how to make reasonable product recommendations for specific groups of people has become one of the most effective ways to expand product revenue. At the same time, in addition to targeted marketing of products, recommendation algorithms are also widely used i...
Examples
Embodiment Construction
[0030] With reference to accompanying drawing, further illustrate the present invention:
[0031] A product recommendation method based on joint non-negative matrix factorization:
[0032] 1. The method comprises the following steps:
[0033] 1) Grab data information from the Internet, including users' ratings on purchased products, friendship between users, and users' text comments on purchased products;
[0034] 2) Transform the data information into a data matrix, and the data information of each user is one of the row vectors;
[0035] 3) Using the method of joint non-negative matrix decomposition, the original data matrix is decomposed into multiple data matrices in low-dimensional space;
[0036] 4) According to the data matrix in the low-dimensional space, estimate the ratings of each user for all unpurchased products, and recommend products according to the ratings.
[0037] The user's rating of the purchased product and the friendship between users described in s...