MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202611109557 A) filed by G. L. Bajaj Institute Of Technology And Management on September 12, 2026, for Adversarially Trained Multilingual Cyberbullying Detection Computing System.
Inventors include Sakshi Yadav; Nishant Tiwari; Prachi Chauhan; and Dr. Harikesh Singh.
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: Disclosed is a computing system for detecting cyberbullying in multilingual or code-mixed social-media text. A network interface receives user-generated text, and a memory stores a multilingual transformer encoder, a classifier, and processing instructions executed by one or more processors. The processors normalize the text to reduce character-level noise and replace informal-language expressions, then encode the normalized text as a contextual representation spanning multiple languages. The classifier is trained using labeled abusive text records and adversarial variants that preserve abusive semantic content while altering surface form. The trained classifier processes the contextual representation to produce a harmful-content indication. Adversarially modified evaluation text is further applied to the trained classifier to assess whether cyberbullying detection remains effective when abusive text is altered by evasive surface-level changes. Dated September 7, 2026 / Kumar Tushar Srivastava IN/PA-3973 Agent for the Applicant
Disclaimer: Curated by HT Syndication.
