Dynamic Category Recommendation via Click Log Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional e-commerce websites face challenges in accurately categorizing products, leading to low return rates due to inaccurate category recommendations, as existing methods rely on preset keyword configurations that do not reflect historical buyer click data.
Innovation Solution
A method and apparatus that calculate correlation information between inquiry words and categories based on stored search click logs, allowing for dynamic matching and categorization by deleting or rephrasing keywords to improve matching accuracy, and using posterior probabilities to select relevant categories.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If preset keyword configuration rules are used to recommend categories, then the system operation is simple, but the category recommendation accuracy is low because it cannot reflect historical buyer click information
Solution Approach 1:
The system uses historical buyer click information as feedback to continuously optimize category recommendations. By analyzing actual buyer behavior data (click logs), the system adjusts and refines category mappings dynamically, ensuring recommendations reflect real user preferences rather than static preset rules.
Solution Approach 2:
The system automatically learns from historical data and performs self-optimization without requiring manual reconfiguration of category rules. The automated analysis of click information enables the system to adapt to changing buyer preferences independently, reducing the need for human intervention in maintaining recommendation accuracy.
2Ease of operation
If hierarchical categorization is used, then the categorization system is structured and manageable, but it is difficult for sellers to choose the correct category when the system is large
Solution Approach 1:
The system introduces an automated intermediary service that bridges sellers and categories. Instead of requiring sellers to manually navigate the hierarchical structure, the system uses automated matching based on product keywords and historical data to directly recommend appropriate categories, simplifying the selection process even in large catalogs.
Solution Approach 2:
The manual mechanical process of browsing and selecting categories through hierarchical menus is replaced with an automated information processing system. The system automatically analyzes product keywords, matches them with historical click data, and computes optimal category recommendations, eliminating the need for manual navigation through large category trees.
3Measurement precision
If text-based keyword matching is used to recommend categories, then the matching process is simple, but it fails when the input keyword does not textually match the category name
Solution Approach 1:
The system changes the matching parameters from simple text string comparison to a multi-dimensional matching approach that incorporates statistical probabilities derived from historical click data. This allows the system to match keywords to categories based on contextual relevance and buyer behavior patterns rather than exact textual overlap.
Solution Approach 2:
The straightforward mechanical text-matching mechanism is replaced with a probabilistic inference system. Instead of relying on exact textual matches, the system uses statistical models that calculate the likelihood of a keyword belonging to a category based on historical data, enabling more flexible and accurate matching.
Data Source
AI summary
The present disclosure provides a method and an apparatus for publishing product information. The present disclosure provides a method for publishing product information. Based on a stored search click log of buyers, correlation information between inquiry words and categories in the search click log is calculated. A keyword input by the seller is matched to the inquiry words. The keyword may be a word or a phrase that includes one or more words. If the keyword is matched to at least one inquiry word, at least one category corresponding to the matched inquiry word is obtained based on the correlation information. The product information is stored under one or more categories of the obtained categories. The present techniques improve the accuracy rate of recommended categories to the seller and the return rate of the published product information.