Device Graph Visitor Identification Beyond Cookie Blocking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional website tracking methods, particularly for unauthenticated visitors, are limited by cookie blocking technologies, impacting web analytics and the ability to offer personalized experiences due to the lack of identifiable identifiers.
Innovation Solution
Utilizing device identifiers such as IP addresses, device fingerprints, and cookies to construct a device graph, and applying community detection algorithms like Louvain or Leiden to determine community information for improved visitor identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If cookies are used for website tracking, then user behavior can be analyzed and users can be identified, but tracking becomes ineffective when cookie blocking technologies are used
Solution Approach 1:
The patent introduces device graphs as an intermediary structure that connects multiple identifiers (cookies, device fingerprints, IP addresses) to represent users. When cookies are blocked, the system uses alternative identifiers linked through the device graph to maintain tracking capability, thus resolving the contradiction between tracking effectiveness and cookie blocking.
Solution Approach 2:
The system dynamically changes identification parameters by switching from relying solely on cookies to using a combination of identifiers including device fingerprints and IP addresses. Community detection algorithms adjust the weight and relevance of different identifiers based on their reliability, allowing the system to adapt to cookie blocking conditions while maintaining tracking effectiveness.
2Reliability
If user registration is required for tracking, then users can be identified reliably, but it limits tracking of users who prefer not to register
Solution Approach 1:
The device graph structure serves multiple functions: it tracks authenticated users through traditional methods while simultaneously tracking unauthenticated users through alternative identifiers like device fingerprints and IP addresses. This universal approach allows the system to maintain reliable identification for registered users while expanding tracking coverage to include users who prefer not to register.
Solution Approach 2:
The system applies partial authentication requirements by allowing tracking of unauthenticated users with lower confidence scores. Community detection algorithms process identifiers even when complete authentication is not present, enabling the system to track a broader audience while maintaining high reliability for fully authenticated users through a tiered approach.
3Measurement precision
If multiple identifiers are collected for device graph construction, then visitor identification accuracy improves, but system complexity increases
Solution Approach 1:
The patent segments the identification system into modular components: identifier collection modules, device graph construction modules, community detection modules, and identification output modules. Each component handles specific tasks independently, making the complex system of multiple identifiers manageable through clear separation of concerns while maintaining high identification accuracy.
Solution Approach 2:
Community detection algorithms provide feedback loops that evaluate the quality and reliability of collected identifiers. The system continuously assesses which identifiers are most effective for identification and adjusts the device graph construction accordingly, optimizing the balance between using multiple identifiers for accuracy and managing system complexity through data-driven decisions.
Data Source
AI summary
Techniques for identifying a visitor device to a website may include capturing device identification information of the visiting device(s). The identification information may include device fingerprints, cookies, domain user identifications and/or other identifiers (e.g., IP addresses, visiting history, etc.). Based on the device identification information, a device graph including a plurality of nodes may be constructed by associating the devices with nodes and connecting the nodes to form the device graph. A community detection algorithm may be applied to the device graph to determine community information for the nodes. Using the identification information and the community information, visitor profiles may be generated, which may be used to identify visiting devices to the website.