Content Selection Using Logistic Regression to Exclude Low-Impact Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing content selection systems fail to effectively identify and mitigate low-impact features that result in lower click-through rates and conversions, leading to inefficient ad placement on web pages.
Innovation Solution
A method and system using a logistic regression model to analyze impression records and determine combination features with low user interest, disabling these features to improve content selection by identifying and excluding them based on weights compared to a threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If content selection systems display all content impressions without filtering, then content volume and coverage are maximized, but click-through rates and conversions decrease due to low-impact features
Solution Approach 1:
The patent extracts and identifies low-impact features from the content selection system using logistic regression analysis. By calculating weights for different features and their combinations, the system isolates features that negatively impact click-through rates and conversions, then excludes them from content selection to improve overall effectiveness
Solution Approach 2:
The system implements feedback through logistic regression modeling that analyzes historical impression data to determine feature weights. This feedback mechanism continuously identifies which features correlate with lower engagement metrics, allowing the system to adaptively exclude low-impact features and improve content selection performance over time
2Productivity
If the system analyzes and excludes low-impact features, then click-through rates and conversions improve, but system complexity increases due to statistical modeling requirements
Solution Approach 1:
The patent replaces complex manual content selection processes with automated logistic regression statistical modeling. This substitution allows the system to objectively and consistently identify low-impact features through mathematical analysis of historical data, reducing the need for manual rule creation and maintenance while improving accuracy
Solution Approach 2:
The system performs self-service by automatically analyzing its own historical impression data to identify which features are impacting performance. The logistic regression model autonomously determines feature weights and identifies low-impact features without requiring external manual analysis, enabling the system to self-optimize its content selection criteria
Data Source
AI summary
Systems and methods of the present disclosure are directed generally to facilitating content selection by identifying low impact criteria. In some implementations, a data processing system accesses a data structure storing, in a memory element, a plurality of impression records. Each impression record can include one or more features and an indication of user interest corresponding to a content impression. The data processing system can identify a combination feature based on at least two of the features. The data processing system can execute a statistical model (e.g., logistic regression model) using the impression records and the combination feature. The data processing system can determine a weight for the combination feature. Responsive to the weight being less than a threshold, the data processing system can transmit an indication to disable the combination feature for selecting content associated with the plurality of impression records.