Media Recommendation Filtering via Location Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users face difficulty in finding media that interests them due to the vast volume of available content, as current recommendation systems struggle to differentiate between intended and incidental captures in visual assets, leading to irrelevant media suggestions.
Innovation Solution
A media guidance application that determines a location score for visual assets captured by users, recommending media only when the score exceeds a threshold, based on user travel patterns, frequency, and popularity of the location, and detects objects within the assets to provide tailored recommendations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the system recommends media based on all visual assets captured by users, then the volume of recommendations increases, but the relevance and quality of recommendations deteriorates due to including incidental captures
Solution Approach 1:
The system performs preliminary analysis of visual assets by detecting objects and determining location information before generating recommendations. It calculates a location score based on user travel patterns and location popularity in advance, filtering out incidental captures before they reach the recommendation stage, thus maintaining high relevance while enabling scalable recommendation volume
Solution Approach 2:
The patent introduces an intermediary filtering mechanism that uses object detection results and location scoring as intermediate steps between capturing visual assets and generating media recommendations. This intermediary layer evaluates whether captured content warrants recommendations by assessing location significance and object presence, thereby separating relevant from incidental captures
2Measurement precision
If the system analyzes user travel patterns and location data to determine location scores, then the accuracy of identifying interesting captures improves, but the computational complexity and processing time increases
Solution Approach 1:
The system pre-calculates and stores user travel patterns, frequently visited locations, and location popularity metrics before they are needed for recommendation. By preparing these data structures in advance, the system reduces real-time computational burden while maintaining high measurement precision in identifying interesting captures
Solution Approach 2:
The system leverages existing mobile device capabilities (GPS, camera, sensors) to automatically collect and process location and visual data without requiring external analytical infrastructure. The device itself performs the heavy lifting of data collection and preliminary processing, reducing the complexity burden on the recommendation server
3Adaptability or versatility
If the system processes location metadata and user behavior data for every visual asset, then the personalization of recommendations improves, but the data processing load and energy consumption increases
Solution Approach 1:
The system applies partial processing by selectively analyzing only certain aspects of visual assets (object detection and location scoring) rather than comprehensively processing all possible data. It uses threshold-based filtering to process only those captures that meet minimum criteria for interest, avoiding unnecessary energy expenditure on clearly incidental captures while maintaining personalization for relevant content
Data Source
AI summary
Systems and methods are disclosed for recommending media assets based on objects captured in visual assets. A location of a visual asset is determined, where the location is associated with the visual asset. A location score is determined based on one or more of: a user travel score, a user travel frequency score, and a popularity score. A determination is made whether the location score exceeds a threshold. Responsive to a determination that the location score exceeds the threshold, the system detects an object within the visual asset and generates a recommendation of a media asset that is associated with the object.