User Profile Generation via Session Event Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing user profiling systems fail to accurately capture short-term and long-term user interests and demographics, leading to ineffective targeting of advertisements, as they do not consider the user's immediate actions and demographics relevant to their interests in various categories.
Innovation Solution
The system infers user profiles by analyzing events from past user sessions, such as web page views and advertisement click-throughs, to determine user interests and demographics from the websites visited, allowing for the generation of both short-term and long-term profiles that can adjust content selection accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If user profiles are generated based on explicit user input, then the profiling process is simple and direct, but the profiles do not accurately capture short-term and long-term user interests and demographics
Solution Approach 1:
The system pre-processes and stores event data from user sessions, categorizing events into vertical categories and computing interest weights in advance. This preliminary action enables the system to quickly generate accurate user profiles when needed, without requiring complex real-time analysis
Solution Approach 2:
The user profile is segmented into multiple dimensions including short-term interests, long-term interests, and demographics. Each dimension is computed separately using different event data and weighting schemes, allowing the system to capture nuanced user characteristics without overwhelming complexity
2Measurement precision
If the system analyzes all user events and behaviors to create comprehensive profiles, then profile accuracy improves, but processing time and computational resources increase
Solution Approach 1:
Event data from user sessions is pre-categorized into vertical categories and stored in an organized structure with pre-computed interest weights. This preliminary processing eliminates the need for time-consuming real-time analysis when generating user profiles
Solution Approach 2:
The system dynamically adjusts the weighting of different event types based on their relevance to user interests. More recent events and higher-engagement events receive greater weights, allowing the system to focus computational resources on the most informative data points
3Measurement precision
If user demographics are inferred from website demographics rather than self-reported data, then demographic accuracy improves, but the system complexity increases
Solution Approach 1:
The system uses website demographics as an intermediary to infer user demographics. Instead of directly asking users for their demographic information or attempting to directly observe user characteristics, the system uses the demographic composition of websites visited by the user as a proxy, which has been shown to correlate with actual user demographics
4Adaptability or versatility
If the system only uses long-term user profiles for content selection, then advertising consistency is maintained, but short-term user interests are not addressed
Solution Approach 1:
The user profile system is segmented into separate short-term and long-term components. The long-term profile captures stable user characteristics and preferences, while the short-term profile captures recent interests and behaviors. Both components are independently computed and then combined for content selection
Solution Approach 2:
The system dynamically combines short-term and long-term profile information based on the specific content selection context. For time-sensitive or trending content, the short-term profile has greater influence, while for evergreen content, the long-term profile dominates
Data Source
AI summary
Methods, systems, and apparatus, including computer program products, for generating user identifier profiles. A user identifier profile characterizes a user based on events that occurred during past user sessions for a user (e.g., past online activities). An event is an action that occurs during a user session, such as a web page view, an advertisement click-through, and a conversion. A user identifier profile can be used, for example, to select advertisements targeted to the user. A user identifier profile includes information about inferred user interests and inferred user demographics.