Dynamic Device Clustering via Authentication Data Merging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for grouping devices associated with users are inefficient due to reliance on common IP addresses, which are restricted by privacy regulations, and are ill-suited for large and dynamic user populations, failing to accurately scale and adapt to real-life anomalies in heterogeneous information networks.
Innovation Solution
A method that starts with a large number of clusters, iteratively merges or adjusts them based on shared devices and user authentication data, using predetermined criteria to ensure each device is associated with a single user, allowing for accurate and scalable clustering without relying on IP addresses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If clustering techniques use predetermined knowledge of the number of clusters (e.g., k-Means algorithms), then the clustering process is simpler and more controlled, but the technique becomes ill-suited for large, unknown, and dynamic user populations
Solution Approach 1:
The patent implements a dynamic clustering approach where the number of clusters is not predetermined but evolves automatically as devices are processed. The system starts with each device as its own cluster and iteratively merges clusters based on shared devices and user authentication data, allowing the cluster structure to adapt dynamically to the actual user population size and composition.
Solution Approach 2:
The system performs preliminary actions by initializing each device as a separate cluster before processing begins. This preliminary setup allows the algorithm to then iteratively merge clusters based on observed sharing patterns, rather than requiring the final number of clusters to be known in advance.
2Device complexity
If clustering techniques assume a very limited number of clusters, then the algorithm is simpler to implement, but the technique does not scale well when the number of clusters is comparable with the number of nodes in the graph
Solution Approach 1:
The patent employs a dynamic clustering methodology where the number of clusters grows or shrinks based on the actual data. Starting with N clusters (one per device), the system iteratively merges clusters that share devices or user authentication data, allowing the final number of clusters to naturally reflect the actual user population rather than being constrained by predetermined limits.
Solution Approach 2:
The system initially segments each device into its own separate cluster, creating a fine-grained initial state. This segmentation allows the algorithm to then progressively merge clusters based on sharing patterns, providing both the granularity needed for accuracy and the flexibility to scale to large numbers of devices and users.
3Ease of manufacture
If existing techniques rely on common IP addresses for device grouping, then the grouping process is straightforward to implement, but the technique becomes restricted by privacy regulations and less accurate
Solution Approach 1:
The patent introduces user authentication data as an intermediary mechanism for device grouping. Instead of directly using IP addresses (which have privacy restrictions), the system uses authentication events that occur on devices as mediating information to infer user-device relationships, thereby achieving accurate grouping while complying with privacy regulations.
Solution Approach 2:
The system replaces the mechanical/IP-address-based grouping mechanism with an authentication-based mechanism. Rather than relying on network layer information (IP addresses), the patent uses application-layer authentication events to determine device ownership, substituting one mechanism with a more privacy-compliant and accurate alternative.
Data Source
AI summary
Techniques are disclosed that provide more accurate clustering of devices by forming clusters of devices and merging or changing clusters based on predetermined criteria. The technique starts with a large number of clusters (e.g., one for each account) and refines the clusters, for example, by merging clusters or determining which cluster a given device should be in when the device is associated with multiple clusters. One technique iteratively adjusts clusters of devices by merging clusters determined to be associated with a single user until a cluster contains all of the devices and accounts expected to be associated with a single user.