A method for hybrid recommendation of consumption places based on visual clustering

Abstract

The invention provides a visual clustering based consumption place mixed recommendation method. The method comprises the steps of: 1) performing data preprocessing on a credit card consumption record; 2) performing RadViz visual clustering on consumption places to discover consumption place clusters with similar consumption features; 3) performing RadViz visual clustering on consumers to discover consumer clusters with similar consumption modes; 4) for the consumers, adding consumption places discovered based on a content recommendation method and a collaborative filtering recommendation method into a corresponding consumer consumption place recommendation list; and 5) further processing the consumer consumption place recommendation list to generate a consumption place recommendation form, thereby competing consumption place personalized recommendation performed on the consumers. According to the method, the personalized recommendation of the most possibly interested consumption places is performed on the consumers from multiple perspectives, and the problems that content based recommendation is only related to consumption places that consumers already go to and collaborative filtering recommendation cannot perform personalized recommendation are solved.

Description

technical field [0001] The present invention relates to consumption location recommendation, especially when there are a large number of credit card consumption records, fully excavating consumers' preferences in certain consumption locations, using a mixed recommendation method, and recommending other consumptions that consumers are most likely to be interested in from multiple perspectives Place. Background technique [0002] Recommender systems have been applied to many fields including search engines, e-commerce, information retrieval, social network services, news media, etc. The current recommendation technologies mainly include: content-based recommendation, collaborative filtering recommendation and association rule-based recommendation. However, these methods have many disadvantages: content-based recommendation is to recommend similar consumption places to consumers based on their existing consumption records. This method can only provide consumers with previously...

AI Technical Summary

Problems solved by technology

However, these methods have many disadvantages: content-based recommendation is to recommend similar consumption places to consumers based on their existing consumption records. This method can only provide consumers with previously interested places, and cannot discover consumption The locations that consumers will be interested in in the future; the recommendation of collaborative filtering is to provide consumers with consumption locations selected by others with similar consumption characteristics. sequence, and consumers have a strict sequence of consumption

[0003] Traditional data mining techniques and algorithms are difficult for users to understand and use, and cannot be user-driven to interact

Images