Hybrid Web Content Classification via Client-Server Rating Combination

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing internet monitor services face challenges in accurately classifying web content, particularly in balancing resource-intensive server-side analysis with user experience on less powerful machines and managing ambiguous content ratings, which can lead to accidental exposure of inappropriate content to users, especially children.

Innovation Solution

An internet monitor service employs a hybrid approach where a monitor client generates a client content rating using real-time classifiers for text and judicious image analysis, while a monitor server provides a server content rating based on large-scale static classifications, with the final rating being a weighted combination of both, allowing for efficient and accurate content assessment.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If server-side analysis is used for content classification, then classification accuracy is improved, but resource consumption and processing time increase

Engineering Contradiction:

Improvecontent classification accuracyVSAvoidresource consumption on client device

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The content classification system is segmented into two parts: a lightweight client-side classifier that performs initial assessment and a server-side classifier that provides comprehensive analysis. The client device executes local code to perform preliminary content rating, while the server performs more resource-intensive analysis only when needed, thus distributing computational load and reducing client resource consumption while maintaining high accuracy.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The client-side classifier performs preliminary content assessment before server-side analysis is triggered. By pre-processing content locally and identifying only ambiguous or potentially problematic content for server review, the system reduces unnecessary server resource consumption and network traffic while maintaining classification accuracy for critical content.

Inventive Principle:

Principle #10Preliminary action

2Measurement precision

If comprehensive content analysis is performed, then content classification accuracy is improved, but user experience on less powerful machines deteriorates

Engineering Contradiction:

Improvecontent classification accuracyVSAvoiduser experience on client device

Core Design Contradiction:

Measurement precisionVSEase of operation

Solution Approach 1:

The classification system is divided between client and server, allowing less powerful machines to run only the lightweight client-side classifier that provides basic protection without consuming excessive local resources. Comprehensive analysis is performed server-side, ensuring high accuracy without burdening client hardware, thus maintaining user experience on diverse devices.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The server acts as an intermediary that handles complex classification tasks. The client device sends content samples to the server for analysis and receives classification results, allowing comprehensive content analysis to be performed without directly impacting client device performance or user experience on less powerful machines.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Speed

If client-side real-time classification is used, then processing speed is improved, but classification accuracy deteriorates due to resource constraints

Engineering Contradiction:

Improvecontent classification speedVSAvoidcontent classification accuracy

Core Design Contradiction:

SpeedVSMeasurement precision

Solution Approach 1:

The system segments classification tasks by urgency and complexity: routine content is classified quickly by the client-side classifier, while ambiguous or high-risk content is flagged for more accurate server-side analysis. This segmentation maintains fast processing for most content while ensuring high accuracy for critical cases through combined client-server classification.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system implements feedback mechanisms where server-side classification results are used to refine and update client-side classifiers over time. Server analysis of ambiguous content provides training data that improves client-side accuracy, creating a feedback loop that progressively enhances client classification precision without sacrificing processing speed.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS9223888B2Combining client and server classifiers to achieve better accuracy and performance results in web page classification

Publication Date: 2015.12.29 MICROSOFT TECHNOLOGY LICENSING LLC

AI summary

In one embodiment, an internet monitor service may use a final content rating to determine access to a webpage. A monitor client 102 may generate a client content rating of a webpage 104. The monitor client 102 may factor the client content rating with a server content rating of the webpage 104 to determine a final content rating for the webpage 104.