Content Recommendation System Using Inverse Popularity Weighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users face difficulty in identifying new content of interest due to the vast variety of content options available, necessitating a system that can recommend content based on individual consumption histories and preferences.
Innovation Solution
A content recommendation system that generates recommendations by analyzing the consumption history of users, identifying similarities in content preferences, and utilizing directional content similarity scores to prioritize less popular content items, allowing users to adjust recommendations based on their preferences and feedback.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If the system recommends content based on popular content items consumed by many users, then the recommendations will be based on sufficient data, but the recommendations will lack personalization and fail to detect subtle user preferences
Solution Approach 1:
The patent changes the parameter of content popularity by introducing inverse popularity weighting. Instead of treating all content equally, the system assigns higher weights to less popular content items that users have consumed, thereby transforming the recommendation approach from popularity-based to preference-based personalization
2Quantity of substance
If the system focuses on commonly consumed popular content items to determine user similarity, then it will have sufficient data to work with, but it will fail to identify users with truly similar content tastes
Solution Approach 1:
The patent transforms the measurement parameter from simple content overlap count to a weighted similarity score that incorporates inverse popularity weights. This allows the system to precisely measure user preference similarity by emphasizing agreements on less popular content items
Solution Approach 2:
The system uses user feedback in the form of consumption history to continuously refine user similarity scores. By incorporating inverse popularity weights based on what users actually consume rather than what is popular, the system adapts to individual user tastes
3Reliability
If the system recommends only popular content items, then it will be easier to find content with sufficient user data, but it will not help users discover new and diverse content
Solution Approach 1:
The patent changes the recommendation parameter by introducing inverse popularity weighting that prioritizes less popular content items. This transformation allows the system to maintain reliability through data-driven approaches while simultaneously increasing content diversity in recommendations
Data Source
AI summary
Disclosed are systems and methods for determining similarities in content preferences among a plurality of users and generating content recommendations based on a requesting user's content consumption history. A requesting user may access the recommendation system to request content items for consumption. The recommendation system may be configured to identify users having consumed the same or similar content items as the requesting user, and to determine a ratio of content items consumed by the requesting user and an identified user to the total number of content items previously consumed by the requesting user. In one embodiment, the system may determine a degree of similarity in content preferences between the two users based on an inverse proportion of a level of popularity associated with the content items consumed by the requesting user and the identified user (e.g., sampled user).