Product Autoclassification Using Category-Specific Probabilistic Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing product classification models in e-commerce are inefficient due to reliance on manual categorization, failure to capture holistic product hierarchies, and lack of guidance for validating category suitability, leading to inaccurate and computationally intensive classification processes.
Innovation Solution
A system utilizing an entity recognizer, model selector, and probabilistic classifiers to select category-specific classification models based on statistical feature models, such as tf-idf and chi-square, for accurate product categorization, reducing computational lag and improving hierarchy validation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual categorization is used to classify products, then precision for single product type is maintained, but productivity decreases and reliability worsens for multiple product types
Solution Approach 1:
The patent segments the classification task by training separate classification models for different product categories rather than using a single universal model. Each category-specific model is trained on category-specific training data, allowing high precision for each product type while automating the overall classification process through multiple specialized classifiers working in parallel.
2Device complexity
If a single classification model is used for all product categories, then device complexity is reduced, but measurement precision decreases across different product types
Solution Approach 1:
The patent applies local quality by making the classification system adaptive to local characteristics of each product category. Each category receives customized training data and a dedicated classification model trained specifically for that category's features and terminology, ensuring high precision for local category requirements while the overall system remains manageable through automated model generation.
3Ease of operation
If conventional models use product title and description as training data, then ease of operation is maintained, but measurement precision decreases due to variability in manual language creation
Solution Approach 1:
The patent incorporates feedback mechanisms where classification results and performance metrics are used to iteratively improve the training data and models. The system evaluates classification accuracy and uses this feedback to refine feature extraction, adjust training data selection, and optimize model parameters, thereby improving precision while maintaining operational simplicity through automated feedback loops.
4Measurement precision
If multiple classification models are trained and tested separately to determine accuracy, then measurement precision is improved, but productivity decreases due to increased computational delay
Solution Approach 1:
The patent applies preliminary action by pre-training and pre-evaluating multiple classification models during the development phase, establishing their relative accuracies in advance. This allows the system to select the most appropriate pre-trained model for each category without performing extensive real-time training and testing, thereby maintaining high accuracy while reducing computational delay during actual product classification operations.
Data Source
AI summary
A system for auto classification of products includes an entity recognizer and a model selector. The entity recognizer receives training data including an attribute of a product. The model selector selects a feature from the training data using a first statistical model to provide a first feature and a second statistical model to provide a second feature, and trains a probabilistic classifier using the first and the second features for providing a first and a second classification models respectively. Further, the model selector calculates an accuracy score of the obtained classification models for each distinct category in a preset hierarchy of categories and selects a classification model from the obtained classification models based on the accuracy score. The selected classification model has a highest accuracy score for a corresponding category in the preset hierarchy.