Cross-Recommendation System Using Merged Behavior Logs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Online shopping malls face limitations in recommending customized products due to restricted access to customer behavior history data across multiple sites, leading to lower accuracy in product recommendations, especially for smaller platforms.
Innovation Solution
A service providing apparatus and method for cross-recommendation between product sales sites, which collects and analyzes behavior log data from multiple sites, calculates category points, matches similarity between sites, and recommends products based on high similarity categories, using collaborative filtering and cosine distance algorithms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If online shopping malls collect behavior history data only from their own websites, then data collection is simple and straightforward, but the variety and accuracy of product recommendations are limited
Solution Approach 1:
The patent merges behavior history data from multiple different online shopping mall websites into a unified analysis system. The server collects behavior logs from various sites, integrates them with product category information, and performs unified analysis to generate cross-site product recommendations, thereby overcoming the limitation of single-site data collection
Solution Approach 2:
The patent introduces a server as an intermediary between multiple online shopping mall websites and users. This server acts as a mediator that collects behavior history data from different sites, processes the information, and provides recommendation services, enabling small and medium-sized shopping malls to access recommendation capabilities without building complex individual systems
2Measurement precision
If small and medium-sized online shopping malls use only their own limited behavior history data, then data collection is easy, but recommendation accuracy is greatly reduced compared to large-scale shopping malls
Solution Approach 1:
The patent combines behavior history data from multiple online shopping mall websites into a unified dataset for analysis. By merging data from different sites, the system enables small and medium-sized shopping malls to leverage collective behavior patterns from multiple sources, achieving recommendation accuracy comparable to large-scale platforms without requiring massive individual data volumes
Solution Approach 2:
The patent creates a universal recommendation system that serves multiple online shopping mall websites simultaneously. The server processes behavior history data from various sites and provides recommendation services to all participating platforms, allowing small and medium-sized shopping malls to access the same recommendation capabilities as large-scale platforms through shared infrastructure
Data Source
AI summary
The present disclosure relates to a service providing apparatus and method for cross-recommendation between product sales sites based on online, a service providing system for the apparatus and method, and a non-transitory computer readable medium having a computer program recorded thereon for calculating and matching similarity between product categories of different product sales sites on the basis of an analysis result according to behavior history analysis of a customer for a plurality of different product sales sites based on online and the apparatus and method recommending a product on the basis of the category of a specific product sales site with high similarity to a recommendation category of another product sales site when the customer visits the specific product sales site on the basis of the matching result.