Web Session Association via Demographic Binary Classifiers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
It is challenging to associate specific web browsing activities with individual members of a household, especially when multiple members use different Internet-accessible devices simultaneously, making it difficult for content providers to gather demographic information accurately.
Innovation Solution
A computer-implemented method that uses binary classifiers trained on demographic attributes to uniquely associate web sessions with household members by processing web events from various devices, employing sessionization and feature extraction to identify distinct user patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple household members use different Internet-accessible devices simultaneously, then the household can access diverse online content and services, but it becomes difficult to associate specific web browsing activities with individual members
Solution Approach 1:
The patent segments the household into individual user profiles with distinct demographic attributes (age, gender, occupation, education). Each user profile is treated as an independent classification target, allowing the system to differentiate between multiple users sharing the same household network. This segmentation enables accurate attribution of web events to specific individuals despite device diversity.
Solution Approach 2:
The patent transforms the classification problem by changing parameters from device-based identification to demographic attribute-based classification. Instead of tracking which device generated which web event, the system uses binary classifiers that analyze web event patterns and match them against demographic parameters of household members. This parameter transformation resolves the ambiguity caused by multiple devices.
2Quantity of substance
If traditional device-based tracking is used, then web browsing activities can be recorded, but it is impossible to determine which household member generated which web events
Solution Approach 1:
The patent introduces demographic attributes as an intermediary layer between web events and user identification. Rather than directly linking web events to users through device identifiers, the system uses demographic characteristics (age, gender, occupation, education) as mediating features that connect browsing patterns to household members. This intermediary approach recovers lost demographic information while preserving web event data.
Solution Approach 2:
The patent performs preliminary classification of web events using trained binary classifiers before final user attribution. The system pre-processes web event data by extracting features and applying demographic-based classification rules, thereby recovering demographic information that would otherwise be lost. This preliminary action enables accurate user identification while maintaining the complete web event dataset.
Data Source
AI summary
A method for associating a web session with a particular member of a group of users includes: receiving a plurality of training web sessions, each training web session including one or more web events generated by a respective known user having one or more demographic attributes; training one or more binary classifiers using the training web sessions and the demographic attributes of the users; receiving a plurality of target web sessions, each target web session including one or more web events that are generated by a respective unknown member of a group of users, wherein each user has one or more demographic attributes; and applying one or more of the binary classifiers to the target web sessions such that a respective target web session is uniquely associated with a member based on, at least in part, the demographic attributes of the member.