Recommendation Model Training with Aggregated Feature Table Reuse
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Constructing recommendation models in existing technologies consumes a large amount of computer resources, leading to low efficiency.
Innovation Solution
Pre-aggregate multiple feature tables corresponding to application scenarios to form an aggregated feature table, which is stored in a cache space, allowing for efficient training of recommendation and neural network models by reusing this table for training.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If feature tables are processed individually for each application scenario, then model training can be performed, but computer resources are wasted and construction efficiency is low
Solution Approach 1:
The patent merges multiple feature tables from different application scenarios into a single unified feature table. This consolidation allows the system to process and store features once rather than repeatedly processing the same features for each scenario, thereby reducing computer resource consumption and improving model construction efficiency across multiple recommendation scenarios.
Solution Approach 2:
The patent performs preliminary aggregation of feature tables into a unified structure before model training begins. By pre-processing and consolidating feature data in advance, the system eliminates the need for repeated feature processing during subsequent model training for different application scenarios, thus reducing resource waste and improving overall efficiency.
2Adaptability or versatility
If multiple feature tables are processed separately for each application scenario, then specific scenario requirements can be met, but the process consumes excessive computer resources
Solution Approach 1:
The patent creates a universal unified feature table that serves multiple application scenarios simultaneously. This single feature table structure is designed to accommodate different recommendation scenarios through flexible feature selection and configuration, allowing the same infrastructure to support diverse application requirements without proportionally increasing resource consumption.
Data Source
AI summary
This application provides a method and apparatus for constructing a recommendation model. In some examples, a plurality of feature tables corresponding to each application scenario in a recommendation project are aggregated to obtain an aggregated feature table. The recommendation project may include a plurality of application scenarios in a one-to-one correspondence with a plurality of recommendation indicators of a to-be-recommended item. Each application scenario can have a recommendation model to predict a corresponding recommendation indicator. Corresponding user feature and item feature may be received from the aggregated feature table based on a user identifier and an item identifier included in a sample data table. The features can be stitched with the sample data table to form a training sample set. The recommendation model of the application scenario may be trained based on the training sample set.