Synthetic Program Records for TV Audience Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for predicting audience performance metrics for new TV programs rely heavily on historical data, which is not available for yet-to-be-released programs, and struggle to account for variations in content characteristics and scheduling.
Innovation Solution
A system that utilizes a machine-learning model trained on a database of historical TV viewing data, where synthetic program records are created by combining historical presentation-logistics features with content-descriptor features from similar programs, and omitting viewer-rating metrics, to predict audience performance metrics for both existing and hypothetical TV programs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods use historical viewer-rating metrics for prediction, then prediction accuracy for existing programs is improved, but prediction cannot be performed for new programs without historical data
Solution Approach 1:
The patent creates synthetic program records that copy the structure and features of historical program records, including presentation-logistics features, content-descriptor features, and simulated viewer-rating metrics. These synthetic records serve as proxies for new programs without historical data, enabling the prediction model to process and analyze new content in the same way as existing content.
Solution Approach 2:
The system performs preliminary actions by generating synthetic program records and training the machine learning model with augmented training data before actual prediction is needed. This preliminary training with synthetic data prepares the model to handle new programs effectively when historical data is unavailable.
2Device complexity
If the system uses only historical program data for training, then model training is simplified, but the system cannot account for variations in content characteristics and scheduling of new programs
Solution Approach 1:
The patent changes key parameters by creating synthetic program records with modified combinations of presentation-logistics features and content-descriptor features. This allows the training data to encompass a broader range of content characteristics and scheduling variations while maintaining the underlying data structure and relationships needed for effective model training.
3Adaptability or versatility
If synthetic program records are created with omitted viewer-rating metrics, then the system can predict metrics for new programs, but the training data becomes less complete
Solution Approach 1:
The patent uses simulated viewer-rating metrics as an intermediary element in synthetic program records. These simulated metrics serve as placeholders that maintain the structural integrity and statistical properties of the training data, allowing the model to learn from complete-looking records while the actual prediction target remains the unknown metric for new programs.
Data Source
AI summary
Methods and systems for prediction audience ratings are disclosed. A database of television (TV) viewing data may include program records for a multiplicity of existing TV programs. A system may receive a training plurality of program records from the TV viewing data, and for each program record a most similar TV program based on content characteristics may be identified. A synthetic program record may be constructed by merging features of each record and its most similar record. Audience performance metrics may be omitted from synthetic records. An aggregate of the training plurality of program records and the synthetic program records may be used to train a machine-learning (ML) model to predict audience performance metrics of the new or hypothetical TV programs not yet available for viewing and/or not yet transmitted or streamed.