Attribute Weighting for Media Content Recommendation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing content recommendation systems suffer from popularity bias, where less popular content is penalized and excluded from recommendations due to poorly trained machine learning algorithms that rely solely on historical user preferences, leading to skewed similarity scores.
Innovation Solution
A hybrid approach that calculates attribute weights based on both historical popularity and objective similarity scores, using a combination of user preference data and content attribute similarity metrics like Cosine similarity or Pearson Correlation, to generate more accurate content-based recommendations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If attribute weights are determined solely based on historical user preferences, then the recommendation system reflects user popularity trends, but popularity bias occurs and less popular content is penalized
Solution Approach 1:
The patent combines two different weighting approaches: (1) user preference-based weights from historical data, and (2) attribute similarity-based weights from content analysis. By merging these two sources of information, the system achieves a balanced weighting scheme that reflects both user preferences and objective content similarity, thereby reducing popularity bias while maintaining recommendation accuracy
Solution Approach 2:
The patent dynamically adjusts attribute weights by introducing a hybrid calculation method that modifies the traditional user-preference-only weighting. The weight for each attribute is changed from a single-source determination to a composite determination involving both user preference metrics and attribute similarity metrics, allowing the system to adapt to both popular and less popular content fairly
2Ease of manufacture
If machine learning algorithms rely solely on historical user preferences, then training is simplified, but similarity scores become skewed and less popular content is excluded
Solution Approach 1:
The patent introduces attribute similarity metrics as an intermediary component that mediates between user preferences and final similarity scores. This intermediary layer provides an objective measure of content similarity that complements user preference data, ensuring that less popular content with high attribute similarity to user favorites can still receive appropriate recommendations
Solution Approach 2:
The patent segments the similarity calculation process into two distinct components: user preference-based similarity and attribute-based similarity. By dividing the overall similarity metric into these two segments, the system can independently optimize each component and combine them to produce more accurate overall similarity scores that don't skew toward only popular content
Data Source
AI summary
Techniques are disclosed for automatically assigning weights to attributes of media content based in part on how many users actually viewed or listened to the content, as well as how many users “liked” or otherwise indicated a preference for the content. The content items can be any type of audio or visual media content, such as songs, videos, or movies, as well as written content, such as books, articles, journals, advertisements, or magazines. A first similarity score is determined based on a similarity between user preferences for content items. A second similarity score is determined based on a similarity between one or more common attributes of the content items. These attributes are assigned ratings that represent the number of users who consumed the corresponding content. Next, weights are assigned to each of the attributes based on the first and second similarity scores using, for example, linear equation regression techniques.