Program Clustering Using Adjusted Viewing Behavior Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing television program clustering methods lack consistency and transparency, relying on subjective content provider inputs and failing to accurately reflect viewer behavior, leading to ineffective genre classification and poor scalability across multiple programs and networks.
Innovation Solution
A behavior-based program clustering system that processes person-level viewing data using advanced statistical techniques to create meaningful clusters based on actual viewing behavior, adjusting for network effects and using metrics like silhouette width to determine optimal cluster numbers, thereby grouping programs by similarity in viewer engagement rather than subjective genre assignments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If subjective content provider inputs are used for program clustering, then the clustering process is simple to implement, but the clustering results lack consistency and transparency
Solution Approach 1:
The patent replaces the mechanical/manual system of subjective content provider inputs with an automated statistical system that uses viewing behavior data and advanced statistical techniques to objectively determine program clusters, thereby improving consistency and transparency while maintaining ease of implementation through automation
Solution Approach 2:
The system enables self-service by allowing viewing behavior data to automatically drive the clustering process without requiring subjective judgment from content providers, with the statistical algorithms independently identifying patterns and creating meaningful clusters based on actual viewer engagement
2Ease of manufacture
If traditional clustering methods are used, then the implementation is straightforward, but the scalability across multiple programs and networks is poor
Solution Approach 1:
The patent creates a universal clustering system that can handle multiple programs and networks simultaneously through standardized statistical methods, allowing the same framework to be applied across diverse content types and platforms, thereby improving scalability while maintaining implementation simplicity
Solution Approach 2:
The system segments the large-scale program clustering problem into manageable components by processing viewing behavior data at person-level and aggregating results across multiple programs and networks, enabling scalable analysis through systematic decomposition of the overall task
3Measurement precision
If person-level viewing data is processed using advanced statistical techniques, then the clustering accuracy reflecting viewer behavior is improved, but the computational complexity increases
Solution Approach 1:
The patent transforms raw person-level viewing data into meaningful clustering parameters through advanced statistical techniques, changing the state of the data from raw observations to processed insights that reveal genuine viewer behavior patterns while managing computational complexity through parameter transformation
Data Source
AI summary
Example apparatus disclosed herein are to compare (i) ratios of program ratings to corresponding network ratings with (ii) a threshold to determine adjusted viewing data for respective sites during a monitoring interval, the program ratings and the corresponding network ratings determined for programs tuned on corresponding networks at the respective sites during the monitoring interval, the adjusted viewing data for a combination of a first program and a first site to represent an adjusted amount of time the first program was presented at the first site. Disclosed example apparatus are also to cluster the programs into program clusters based on distances between respective combinations of pairs of the programs, the distances based on the adjusted viewing data. Disclosed example apparatus are further to output information to identify the program clusters.