Multi-angle hybrid recommendation method for cosmetics

Abstract

The invention relates to a multi-angle hybrid recommendation method for cosmetics. The method comprises the following steps that: 1) user data are acquired, wherein the user data include the IP addresses of users, and therefore, the geographic locations, current seasons and cosmetic score information of the users can be derived according to the IP addresses; 2) a skin nature model is built according to the user data; 3) singular value decomposition pre-processing is performed on a matrix requiring similarity calculation; 4) the set S of cosmetics which the users do not score is obtained from a user-score matrix; 5) the assessed values of the cosmetics are given according to the skin natures of the users, the popularity of the cosmetics, the tags of the cosmetics and mapping relations of the skin natures and tags; and 6) the cosmetics in the set S of cosmetics which the users do not score are sequenced according to the high-to-low order of the assessed values of the cosmetics, and sequence results are adopted as recommendation results, and the recommendation results are displayed for the users sequentially. With the method of the invention adopted, personalized recommendation can be realized, and the user experience of the users in cosmetic selection can be improved.

Description

technical field [0001] The invention relates to the field of information technology, in particular to a multi-angle mixed recommendation method for cosmetics. Background technique [0002] With the development of Internet technology, people are in an era of information explosion. In the face of massive information data at this stage, certain rules are needed to screen and filter massive data for users and present the information that users are most interested in in front of users, saving It shortens the time for users to screen information and increases the user experience at the same time. [0003] The emergence of search engines has solved the problem of information screening to a certain extent, but it cannot actively provide information for users. At the same time, the information can be actively displayed among the people who are interested in it. In the cosmetics industry, the types and effects of cosmetics are complex, and users may not know much about the products ...

AI Technical Summary

Problems solved by technology

[0004] In order to overcome the shortcomings that the current recommendation system cannot make personalized recommendations for factors such as cosmetic characteristics, user skin attributes, cosmetic popularity, and user environmental characteristics, the present invention provides a multi-angle mixed recommendation method for cosmetics, specifically Said that it is a multi-dimensional information modeling of cosmetics users, mixing multiple recommendation algorithms, using the user's ratings on cosmetics, cosmetic labels, user skin attributes, and the mapping relationship between skin attributes and cosmetic labels to achieve combined user Preferences, user skin characteristics and current popular products are a system for the purpose of personalized recommendation of cosmetics for users, so as to improve user experience in cosmetics selection

Images