Weighted Directed Graph User Correlation for E-commerce Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing user clustering methods in e-commerce platforms are inefficient due to variations in user access patterns, leading to poor clustering effects and reduced sales conversion rates, as they struggle to accurately measure user correlation and provide personalized recommendations.
Innovation Solution
A method and apparatus that utilize clickstream data to compute click path correlations between users by constructing weighted-directed graphs, assigning hierarchical and proportional weights based on webpage visits, and configuring comprehensive weights for preset tags, to identify user similarities and provide personalized recommendations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If user clustering is based on sequence of webpage hits from user access paths, then user behavior data can be collected, but the clustering effect is poor due to variations in user access patterns
Solution Approach 1:
The patent transforms the user access path sequence into a weighted graph structure, changing the parameter representation from linear sequence to network topology. This allows capturing user behavior patterns while eliminating the negative impact of access sequence variations, thereby improving both measurement accuracy and clustering reliability
Solution Approach 2:
The patent segments the user access path into individual webpage nodes and their relationships, representing each page as a node and navigation as edges. This segmentation allows the system to focus on structural patterns rather than sequential patterns, resolving the contradiction between data collection and clustering effectiveness
2Quantity of substance
If multiple user visits are aggregated, then more behavior data is available, but the differences caused by multiple visits reduce clustering accuracy
Solution Approach 1:
The patent merges multiple user visits by constructing a unified weighted graph that aggregates navigation patterns across visits. Instead of treating each visit as a separate sequence, the system combines them into a single structural representation where edge weights reflect aggregated navigation frequencies, preserving data volume while eliminating visit-to-visit variations
3Ease of manufacture
If traditional user clustering is used, then implementation is simple, but sales conversion rate is reduced due to poor clustering effect
Solution Approach 1:
The patent replaces traditional sequence-based clustering mechanisms with a graph theory-based system. By substituting linear sequence processing with graph structure analysis, the system achieves more accurate user correlation measurement while maintaining computational feasibility through established graph algorithms
Data Source
Figure 1
Figure 2~3
Figure 4~5
AI summary
A method for extracting user features based on user behaviors. The method uses webpage clickstream data of a current user to compute a path correlation between the current user and other users, selects a number of other users whose path correlation with the current user ranks among the highest, and then configures weights in connection to preset tags of websites visited by the selected other users, and computes a user correlation between the current user and the selected other users based on the preset tags and the weights. The method constructs weighted-directed graphs of webpage click paths based on click stream data, and converts computing user correlation to computing a similarity of weighted-directed graphs. The method further combines computing correlation of webpage tags to discover the user's clicking habits and personal preferences, and improve the accuracy and efficiency of user clustering.