User Behavior Prediction Model Using Interaction Contribution Values
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing user behavior prediction methods, such as those using linear regression models, suffer from low accuracy as they only consider the individual impact of characteristic data without accounting for interactions between different categories of data, leading to poor prediction results and high calculation complexity.
Innovation Solution
A method that processes groups of characteristic data using specific interaction models corresponding to their categories, calculating both individual and interaction contribution values to determine the execution probability of a behavior, thereby improving prediction accuracy while reducing complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a linear regression model is used to predict user behavior by calculating individual contribution values of characteristic data, then the calculation process is simple, but the prediction accuracy is relatively low
Solution Approach 1:
The patent segments the characteristic data into different categories (first category, second category, third category, etc.) and processes each category separately using category-specific contribution models. This segmentation allows the system to capture category-specific interaction patterns while maintaining manageable computational complexity through modular processing.
Solution Approach 2:
The patent combines multiple types of contribution values (individual contribution values from linear regression, interaction contribution values from category-specific models, and cross-category interaction contribution values) to form a composite prediction model. This composite approach integrates simple and complex modeling techniques to achieve both computational efficiency and high prediction accuracy.
2Measurement precision
If interaction models are applied to every group of characteristic data to capture interaction impacts, then prediction accuracy improves, but calculation complexity increases excessively
Solution Approach 1:
The patent divides characteristic data into distinct categories and applies category-specific contribution models only within each category rather than across all possible data combinations. This segmentation dramatically reduces the number of interaction models needed while still capturing essential interaction effects within each category.
Solution Approach 2:
The patent uses a universal linear regression model to calculate individual contribution values for all characteristic data, and then supplements this with category-specific interaction models only where needed. This multi-functional approach applies different modeling complexities selectively, maintaining efficiency while improving accuracy.
3Device complexity
If a single characteristic interaction model is used to process all characteristic data, then calculation complexity is reduced, but prediction results become poor due to inability to capture category-specific interactions
Solution Approach 1:
The patent assigns different quality levels and modeling approaches to different categories of characteristic data. Each category receives a dedicated contribution model tailored to its specific characteristics and interaction patterns, ensuring that locally optimal modeling strategies are applied where they are most needed.
Solution Approach 2:
The patent segments characteristic data into multiple categories (first, second, third, fourth categories) and processes each segment with appropriate interaction models. This segmentation enables the system to capture category-specific interaction patterns that a single universal model would miss.
Data Source
AI summary
Example user behavior prediction methods and apparatus are described. One example method includes obtaining a first contribution value of each piece of characteristic data for a specified behavior after obtaining behavior prediction information including a plurality of pieces of characteristic data. Every N pieces of characteristic data in the plurality of pieces of characteristic data may be processed by using one corresponding characteristic interaction model, to obtain a second contribution value of the every N pieces of characteristic data for the specified behavior. Finally, an execution probability of executing the specified behavior by a user may be determined based on the obtained first contribution value and the obtained second contribution value, to predict a user behavior. In the example method, interaction impact of the plurality of pieces of characteristic data on the specified behavior is considered during behavior prediction.