Session Group Profiles for Media Recommendation Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Content recommendation systems often fail to accurately suggest media content to users due to shared accounts and diverse user demographics, leading to irrelevant recommendations based on content similarity rather than contextual viewing sessions.
Innovation Solution
A method that groups viewing sessions by session attributes such as application context, device configuration, and shared session information to create session group profiles, allowing for personalized content recommendations by matching current viewing sessions with relevant profiles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If content recommendation systems use shared user accounts to aggregate viewing data, then the system can provide personalized recommendations, but the recommendations become irrelevant due to diverse user demographics and consumption patterns
Solution Approach 1:
The patent segments the unified user account into multiple persona profiles (e.g., primary user, family members, guests) based on viewing session attributes. Each persona maintains separate viewing history and preferences, allowing the system to provide personalized recommendations for each individual while using a shared account. This resolves the contradiction by dividing the monolithic account data structure into segmented persona-specific segments.
Solution Approach 2:
The patent introduces persona profiles as intermediary entities between the shared user account and the recommendation engine. These personas act as mediators that attribute viewing sessions to specific users based on device, time, and content preferences, enabling accurate recommendations without requiring separate account logins. The intermediary persona layer filters and organizes the diverse consumption patterns from multiple users.
2Quantity of substance
If the system recommends content based on all content consumed in a user account, then it can leverage extensive viewing data, but the recommendations include content from other users that the authenticated user is not actually interested in
Solution Approach 1:
The patent segments the aggregate viewing data by attributing each viewing session to a specific persona profile based on session attributes like device type, time of day, and content genre preferences. This segmentation filters out content consumed by other users or in different contexts, retaining only the relevant viewing history for the authenticated user while preserving the benefits of extensive data collection.
Solution Approach 2:
The patent applies local quality by creating persona-specific viewing histories with different weights and priorities. Each persona has its own quality-adjusted view of the content library, where content preferences and viewing patterns are tailored to that specific persona's demonstrated interests rather than treating all users uniformly.
3Measurement precision
If the system uses session attribute information to group viewing sessions, then it can create contextually relevant recommendations, but the system complexity increases due to multiple session attributes and grouping criteria
Solution Approach 1:
The patent segments session attributes into hierarchical categories (e.g., device type, time period, content genre) and applies weighted scoring to each category. This segmented approach to attribute processing reduces complexity by organizing diverse session attributes into manageable segments with predefined importance weights, making the grouping process more systematic and less computationally intensive.
Solution Approach 2:
The patent dynamically adjusts the weight and threshold parameters for different session attributes based on user behavior patterns. For example, device type may be weighted more heavily for certain users while time of day is more important for others. This parameter adaptation allows the system to maintain high contextual relevance while simplifying the decision logic by adjusting which attributes matter most in different contexts.
Data Source
AI summary
In some embodiments, a method for recommending content comprises: receiving an authorization to access a media content consumption history, wherein the media content consumption history includes a plurality of viewing sessions that each include media content items that have been consumed during that viewing session and wherein each of the plurality of viewing sessions is associated with session attribute information; generating a plurality of session group profiles by grouping a subset of viewing sessions from the plurality of viewing sessions based on the session attribute information; determining that a user device is consuming a media content item in a current viewing session; in response to determining that the current viewing session matches a session group profile from the plurality of session group profiles, and causing a recommended media content item to be presented on the user device.