Member shopping data mining algorithm comprehensive engine

Abstract

The most mainstream algorithm in the current e-commerce website is a recommendation algorithm based on cooperation in combination with a recommendation algorithm based on content. However, the algorithms all have the problems that new users or new commodities are subjected to cold start, and a large amount of historical data is needed. Especially, in the supermarket shopping process, it is difficult to force or guide users to log in for shopping, and great deviation can be brought to the processing result when the users who do not log in are processed as new users. Most of the existing algorithms aim at the e-commerce state, access is allowed compared with the characteristics of supermarkets, and gaps exist between the existing algorithms and supermarket shopping recommendation scenes. According to business characteristics of a supermarket shopping scene, supermarket purchaser composition and analysis of behavior characteristics of the supermarket purchaser are combined, and the invention provides a hybrid recommendation method for a supermarket shopping scene. According to the method, a recommendation method based on CF (Collaborative Filtering) is taken as a main method, a recommendation method based on CB (Content-based) is taken as an auxiliary method, and a feature supplementary hybrid recommendation algorithm of a specific user portrait map is combined.

Description

technical field [0001] This engine provides a data analysis algorithm for members combined with products, specifically involving members, products, promotions, and algorithms for members' shopping preferences after staying on the shelf, which belongs to the field of data structure and algorithms. Background technique [0002] At present, in the era of big data, the styles of online platforms and consumers’ shopping habits are diversified, and the collection of consumer data and the analysis of behaviors need to be gradually expanded to more data sources, combined with shopping websites, other web browsing information, social media platforms, etc. Information, mobile terminals, search engines and other platforms to contact consumers and mine data for comprehensive evaluation and analysis. The most mainstream algorithms in current e-commerce websites are recommendation algorithms based on collaboration combined with recommendation algorithms based on content. However, these a...

AI Technical Summary

Problems solved by technology

Its beneficial effect is that it can solve the cold start problem of some new users for supermarket shopping scenarios, and add supplementary features for the lack of user evaluation, optimize the product recommendation algorithm, and improve the accuracy of recommendation to a certain extent

Its beneficial effect is that it can divide more accurate user portraits for supermarket shopping scenarios, and provides a certain possibility, so that users can timely adjust the partial output results of the user's user portraits during the shopping process. Reduce the problems caused by cold start of new users to a certain extent

Images