Media Content Filtering via Classification Certificates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Managing the distribution of media content in online social networking services is challenging due to the presence of inappropriate, malicious, or spam content, which varies by user and requires significant computing resources for classification, often delaying access to content.
Innovation Solution
A content filtering system that classifies and filters media content using classification certificates, applying varying filtering rules based on user preferences, regulatory requirements, and content properties, filtering at both ingestion and retrieval times to ensure appropriate content distribution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If media content is classified using significant computing resources, then classification accuracy improves, but content access time increases
Solution Approach 1:
The system performs preliminary classification of media content at the time of upload or ingestion, generating classification results in advance before the content is requested by users. This preliminary action stores classification data that can be quickly retrieved later, eliminating the need for time-consuming classification processing when users access the content.
Solution Approach 2:
The system creates and stores copies of classification results separately from the original media content. These classification copies contain the essential classification information that can be quickly accessed and applied when content is requested, avoiding the need to re-perform the entire classification process during content access.
2Device complexity
If uniform filtering rules are applied to all users, then system simplicity is maintained, but user satisfaction decreases due to inappropriate content restrictions
Solution Approach 1:
The system applies different filtering rules to different users based on their individual characteristics, preferences, and requirements. Each user receives customized content filtering that matches their specific needs, rather than applying a single uniform set of rules to all users. This local differentiation improves user satisfaction while maintaining manageable system complexity through modular rule application.
Solution Approach 2:
The filtering rules are made dynamic and adaptable rather than static and fixed. The system can adjust filtering criteria based on user feedback, changing requirements, or contextual factors, allowing the filtering behavior to evolve and optimize for each user's satisfaction while keeping the underlying system structure relatively simple.
3Speed
If all media content is made visible immediately, then user access speed improves, but exposure to inappropriate content increases
Solution Approach 1:
The system performs preliminary filtering and classification of media content at the time of upload, identifying and flagging inappropriate content before it can be exposed to users. This advance preparation allows the system to quickly serve appropriate content while blocking harmful material, achieving both fast access speed and content safety.
Solution Approach 2:
The system introduces an intermediary filtering layer between content storage and user access. This intermediary component automatically screens content based on classification results and filtering rules, allowing appropriate content to pass through quickly to users while blocking inappropriate content, thus mediating between speed and safety requirements.
Data Source
AI summary
This disclosure relates to systems and methods that include receiving media content from a content submitter, classifying the media content by initiating one of synchronous classification and asynchronous classification based on a type of the media content, generating a media content certificate based on the media type, the certificate including results of the classification, storing the media content certificate with the media content, and filtering the media content based on at least one of an identity of the content submitter, the results of the classification, and the media content including malicious content.