Pattern-based classification
Topics: Probably in use, User Signals
This Google patent describes a method for classifying sequences of user interactions with digital content as either valid or invalid. The system analyzes interaction data including event types and time delays between events, encodes this data into a standardized format, and uses a trained model to determine if the interaction pattern is legitimate or suspicious. Based on this classification, the system can prevent content distribution to entities showing invalid interaction patterns, thereby conserving computing resources and preventing content delivery to automated bots or compromised devices. This technology can have serious impact on SEO tactis like CTR manipulation.