Patent Application for Toxicity Detection in Virtual Environments Using Natural Language Processing

Strategic Insights -

A team of inventors, including Chakra, Al; Chung, David-Jude; Duzenli, Emre; and Fazzone, Davide, has filed a patent application for a system, method, and computer program product to detect and mitigate inappropriate behavior in virtual environments. The system uses natural language processing to analyze discussions between users and calculate a toxicity score based on the presence of certain words or phrases. The inventors claim that their system is more effective than current methods, which often rely on human moderators or simple keyword filtering.

Key Takeaways:

  • The patent application, filed on January 4, 2024, describes a system that uses natural language processing to detect toxicity in virtual environments, including online discussions, chat rooms, and virtual reality experiences.
  • The system analyzes discussions for toxicity indicators, such as abusive language, profanity, and insults, and assigns a toxicity score to each user based on the presence of these indicators.
  • The system can also take into account factors such as user familiarity, age, and location to adjust the toxicity score.
  • The inventors claim that their system is more effective than current methods, which often rely on human moderators or simple keyword filtering, as it can analyze complex language patterns and context.
  • The system includes a range of operations, including detecting discussions, analyzing toxicity indicators, calculating toxicity scores, and implementing responses to high-scoring users.
  • The inventors also describe a method of correlating toxicity indicators with events in virtual environments, and predicting toxic reactions to certain events.
  • A computer program product is also described, which includes a computer-readable storage medium with program instructions to perform the functions of the system.

Statistics:

  • The patent application includes 20 claims, which cover a range of inventions related to toxicity detection and mitigation in virtual environments.
  • The inventors claim that their system can detect toxicity in virtual environments with a high degree of accuracy, which is not possible with current methods.
  • The system can analyze large volumes of data, including online discussions and chat logs, to identify patterns of toxicity.
  • The inventors also claim that their system can adapt to changing language patterns and context, making it more effective than current methods.

Sources:

  • Chakra, Al; Chung, David-Jude; Duzenli, Emre; Fazzone, Davide. Toxicity Detection Using Natural Language Processing. U.S. Patent Application Number 20250225324, filed January 4, 2024 and posted July 10, 2025. Patent URL (for desktop use only): https://ppubs.uspto.gov/pubwebapp/external.html?q=(20250225324)&db=US-PGPUB&type=ids