Hybrid Recommendation Engine Using Segmented User Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Website publishers face challenges in efficiently filtering through vast amounts of user information to accurately predict and present targeted content, leading to increased computing costs and diminished user experience due to processing time.
Innovation Solution
Combining selection rules and machine learning techniques, such as support vector machines and factorization machines, to create predictors that select users likely to engage with specific content, while adapting strategies like Multi-Armed Bandit to improve performance over time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If filtering through vast amounts of user information is performed to accurately predict targeted content, then prediction accuracy is improved, but processing time and computing costs increase
Solution Approach 1:
The patent segments the immense user information into structured data formats with defined schemas, organizing data into manageable components that can be processed more efficiently by machine learning models, thereby reducing processing time while maintaining prediction accuracy
Solution Approach 2:
The patent transforms unstructured or semi-structured user information into structured data with specific parameters and schemas, changing the data format to enable more efficient processing by predictive models, thus resolving the contradiction between comprehensive analysis and processing speed
2Measurement precision
If filtering through vast amounts of user information is performed to accurately predict targeted content, then prediction accuracy is improved, but computing costs increase
Solution Approach 1:
The patent segments user information into structured data components with defined schemas, allowing predictive models to process only relevant structured features rather than raw unstructured data, thereby reducing computing costs while maintaining prediction accuracy
Solution Approach 2:
The patent creates structured data representations (copies) of user information that preserve essential predictive features while reducing data complexity, enabling efficient processing by machine learning models without analyzing the entire raw dataset
3Reliability
If extensive analysis of user information is performed, then targeted content prediction is improved, but user experience deteriorates due to processing time
Solution Approach 1:
The patent performs preliminary structuring and organization of user information data before the prediction process, preparing data in advance with defined schemas so that when prediction is needed, the structured data can be quickly processed by machine learning models, reducing wait time for users while maintaining prediction reliability
Data Source
AI summary
To maximize the accuracy and efficiency of predicting users that will enjoy targeted content, a proposed content selection solution looks to combine a first strategy of utilizing selection rules with a second strategy of utilizing machine based learning models. By combining the selection rules-based approach and the machine learning model-based approach, the proposed content selection solution is able to consider and recommend a wider range of users for each available content.