Feature Construction for Content Display Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing content display technologies only consider user information and content information, resulting in limited analysis dimensions and inefficient resource utilization, leading to waste in content display and reduced user conversion rates.
Innovation Solution
A method for feature construction that acquires interaction data and loading performance data to construct user interaction features and page performance features, which are used to train a content display model to determine target content for users, improving accuracy and resource utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If only user information and content information are considered for content display analysis, then the analysis process is simple, but the analysis dimension is limited and content display accuracy is reduced
Solution Approach 1:
The patent introduces a new dimension of analysis by incorporating page loading performance data (loading speed, rendering time, resource loading status) alongside traditional user information and content information. This multi-dimensional feature construction approach expands the analysis space from 2D (user+content) to 3D (user+content+performance), enabling more comprehensive content display optimization without excessively increasing system complexity
2Measurement precision
If more data dimensions are incorporated into content display analysis, then content display accuracy is improved, but resource consumption increases
Solution Approach 1:
The patent performs preliminary analysis by pre-processing and storing page loading performance data during normal operation. This pre-computed performance information is then readily available for content display decisions, avoiding the need for real-time heavy computation when actual content matching occurs, thus reducing instantaneous resource consumption while maintaining high accuracy
Solution Approach 2:
The patent creates feature representations (copies) of complex multi-dimensional data including user behavior patterns, content characteristics, and page performance metrics. These compressed feature vectors serve as efficient proxies for the original detailed data, enabling accurate content matching with reduced computational overhead and resource usage during inference
3Productivity
If traditional user and content information only are used, then data processing is efficient, but user conversion rate is reduced due to wasted content display resources
Solution Approach 1:
The patent transforms the content display problem by changing key parameters: instead of only considering user preferences and content attributes, it incorporates page loading performance parameters (loading speed, rendering time, resource status) as new decision variables. This parameter expansion enables more precise content-ranking-page matching, improving user conversion rates while maintaining processing efficiency through feature-based modeling
Data Source
AI summary
The present disclosure relates to a method for feature construction, a method for content display and a related apparatus. The method for feature construction comprises: acquiring interaction data on a content page and loading performance data of the content page; constructing a user interaction feature according to the interaction data on the content page, and constructing a page performance feature of the content page according to the loading performance data of the content page. The user interaction feature and the page performance feature are used for training a content display model, and the content display model is used for determining target content displayed to a target user.