Media Recommendation Accuracy via Multi-Source Data Normalization

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Traditional media asset recommendation systems rely on a single data set from a single content provider, limiting their accuracy and failing to fully leverage diverse user preference information across multiple data spaces.

Innovation Solution

The system determines error values for training models by comparing expected media asset similarity values from multiple data spaces, normalizing user preference information, and adjusting trainable parameters to improve recommendation accuracy by aggregating data from various content providers using different metrics.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional systems use a single data set from a single content provider to determine media asset recommendations, then the system complexity is low and ease of operation is maintained, but the recommendation accuracy is limited

Engineering Contradiction:

Improverecommendation accuracyVSAvoidsystem complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent combines multiple data sets from different content providers into a unified recommendation system. The system aggregates user preference information across multiple data spaces, merging diverse data sources to improve recommendation accuracy while managing the complexity through structured integration approaches.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The recommendation system is designed to handle multiple data sources and different types of user preference information universally. It can process various metrics from different content providers and adapt to diverse data formats, making the system multi-functional in handling different data spaces while maintaining a unified recommendation framework.

Inventive Principle:

Principle #6Universality (Multi-functionality)

2Measurement precision

If the system aggregates user preference information from multiple data spaces with different metrics, then the recommendation accuracy improves, but the difficulty of detecting and measuring increases

Engineering Contradiction:

Improveuser preference measurement accuracyVSAvoiddata normalization difficulty

Core Design Contradiction:

Measurement precisionVSDifficulty of detecting and measuring

Solution Approach 1:

The system applies parameter changes by normalizing different metrics from multiple data spaces into a common framework. It transforms various user preference measurements (watch time, ratings, engagement metrics) from different content providers into standardized parameters that can be consistently compared and processed, thereby improving measurement accuracy while managing the complexity of diverse data sources.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS10003836B2Systems and methods for improving accuracy in media asset recommendation models based on users' levels of enjoyment with respect to media assets

Publication Date: 2018.06.19 ADEIA GUIDES INC

AI summary

Methods and systems for determining an error value based on the user's expected level of enjoyment with respect to a specific media asset based on the user's level of enjoyment of other media assets, as determined using a model. User preference information is received from two data spaces that are managed by different content providers. User preference information from the two data spaces is normalized and an indication of similarity between two media assets is determined. The user's level of enjoyment with respect to a media asset is compared to an expected user's level of enjoyment with respect to the media asset received from a model and an error value is determined based on the comparison.