Machine Learning Risk Score for E-Commerce Item Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing e-commerce systems fail to effectively detect and remove risky items, such as weapons and hazardous materials, as sellers evade detection by avoiding specific keywords on item pages.
Innovation Solution
A machine learning model is trained to determine the likelihood of an item page representing a risky item by analyzing item textual descriptions and search queries, computing probability scores for topics associated with risky items, and generating cumulative and total risk scores to identify potentially hazardous products.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If keyword-based detection is used to identify risky items, then the detection process is simple and fast, but sellers can easily evade detection by avoiding specific keywords
Solution Approach 1:
The patent replaces the mechanical keyword-matching system with a machine learning-based image recognition system. Instead of relying on text-based keyword detection that sellers can evade, the system uses computer vision algorithms to automatically analyze product images and identify risky items based on visual characteristics, thereby improving detection accuracy while maintaining automated processing speed
Solution Approach 2:
The patent changes the detection parameter from text-based keywords to image-based visual features. By transforming the detection approach from analyzing textual descriptions to analyzing visual characteristics of product images, the system overcomes the limitation of keyword evasion while enabling more reliable identification of risky items
2Measurement precision
If manual review of item pages is performed to detect risky items, then detection accuracy improves, but the processing time and operational complexity increase significantly
Solution Approach 1:
The patent implements an automated machine learning system that performs detection without requiring manual human review. The system serves itself by automatically analyzing product images, generating risk scores, and identifying risky items, thereby maintaining high detection accuracy while eliminating the time loss and operational complexity associated with manual processing
Solution Approach 2:
The patent substitutes manual human review with an automated machine learning-based image analysis system. This replacement maintains or improves detection accuracy through consistent application of detection algorithms while dramatically reducing processing time by eliminating human intervention in the detection process
Data Source
AI summary
Disclosed are systems and methods that determine a likelihood that an item page represents a risky item. For example, a machine learning model may be trained to determine a probability score that an item page corresponds to a topic of a plurality of topics based on item textual descriptions determined from the item page and search queries corresponding to the item page. Still further, one or more of a topic risk score, item risk score, cumulative item risk score, and/or a total item risk score may be determined. Each of the risk scores may be indicative of a likelihood that the item page represents a risky item.