Multimodal Neural Graph for Automated Gender Label Assignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Assigning gender attributes to items in a catalog database is a time-consuming manual process, and existing technologies lack efficient methods for automatically determining gender labels for items without pre-defined attributes.
Innovation Solution
A multimodal neural graph system that uses text embeddings and machine learning models to automatically assign gender labels by analyzing item descriptions and co-view data from user sessions, propagating labels through a graph structure based on co-view counts, and training models with neural loss functions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual process is used to assign gender attribute values to items in a catalog database, then accuracy of gender attribute assignment can be maintained, but time consumption and processing efficiency deteriorate
Solution Approach 1:
The patent replaces the manual mechanical process of gender attribute assignment with an automated machine learning system. The system uses trained models that analyze item attributes, descriptions, and co-view patterns to automatically predict and assign gender attributes, eliminating the need for manual human intervention while maintaining high accuracy through sophisticated algorithms and large-scale training data
Solution Approach 2:
The system enables the catalog database to self-assign gender attributes through automated machine learning models. The models continuously learn from item data and co-view patterns, allowing the system to autonomously populate gender attributes for new and existing items without requiring manual input, thereby achieving both speed and accuracy
2Productivity
If automated methods are used to assign gender attributes to items, then processing efficiency and speed are improved, but accuracy and reliability of gender attribute assignment deteriorate
Solution Approach 1:
The patent implements preliminary training of machine learning models using extensive labeled data and co-view patterns before deployment. This pre-training phase establishes accurate prediction capabilities that enable the system to reliably assign gender attributes at high speed during actual operation, ensuring both efficiency and accuracy are achieved
Solution Approach 2:
The system incorporates feedback mechanisms where model predictions are continuously refined based on actual user interactions and co-view data. This feedback loop allows the model to learn from real-world usage patterns, improving accuracy over time while maintaining high processing efficiency through automated updates
3Measurement precision
If comprehensive item analysis is performed to determine gender attributes, then accuracy of gender label assignment is improved, but system complexity and computational resources required deteriorate
Solution Approach 1:
The patent segments the gender attribute assignment system into distinct functional modules: feature extraction from item attributes, co-view pattern analysis, machine learning model prediction, and result integration. This modular segmentation reduces system complexity by making each component independent and manageable while maintaining comprehensive analysis capabilities for accurate gender label assignment
Data Source
AI summary
A system including one or more processors and one or more non-transitory computer-readable media storing computing instructions that, when executed on the one or more processors, cause the one or more processors to perform functions including: receiving a respective item description and at least one respective attribute value for each item of a set of items; generating at least one respective text embedding; generating a graph of the set of items based on at least co-view data to create pairs of items that are co-viewed by joining respective pairs of items; training the text embedding model and a machine learning model using a neural loss function based on the graph; and automatically determining, using the machine learning model, as trained, a label for each item of the set of items. Other embodiments are disclosed.