Personalized service recommendation method and system based on data portrait and fusion algorithm

CN122309834APending Publication Date: 2026-06-30HUNAN GREATWALL INFORMATION FINANCIAL EQUIP

Patent Information

Authority / Receiving Office: CN · China
Patent Type: Applications(China)
Current Assignee / Owner: HUNAN GREATWALL INFORMATION FINANCIAL EQUIP
Filing Date: 2026-02-24
Publication Date: 2026-06-30

AI Technical Summary

Technical Problem

Traditional banking recommendation systems struggle to accurately predict user behavior and provide personalized recommendations, and are often based on a single data source or a single recommendation algorithm, failing to fully capture the complexity of user needs and business scenarios.

Method used

By deeply mining user data and branch data profiles, combining the FP-Growth algorithm for association rule mining and the k-means algorithm for clustering, and combining item-based collaborative filtering and content-based recommendation algorithms, a personalized recommendation list is generated. The logistic regression model is then used for fusion optimization, and finally, personalized business recommendations are provided on self-service devices.

Benefits of technology

This improved the accuracy, relevance, and usability of business recommendations, ensuring that the recommendations matched the actual business functions offered by the branches, thereby enhancing user experience and improving the efficiency of banking transactions.

✦ Generated by Eureka AI based on patent content.

Smart Images

Figure CN122309834A_ABST

Patent Text Reader

Abstract

This invention discloses a personalized business recommendation method and system based on data profiling and fusion algorithms. The method includes: acquiring and preprocessing user profiles, branch profiles, and business function data; obtaining a set of association rules through association rule mining algorithms, and clustering user groups based on user and branch features; for target users, executing item-based collaborative filtering and content-based recommendation algorithms in parallel to generate a first recommendation list and a second recommendation list, respectively; inputting the scores of the two lists into a pre-trained fusion model for weighted fusion to obtain personalized recommendation results; finally, filtering based on the scope of services that can be handled at the current branch, and pushing the final recommendation list to self-service devices to achieve accurate personalized recommendations and improve user experience and business processing efficiency.

Need to check novelty before this filing date? Find Prior Art