MUMBAI, India, Sept. 14 -- Intellectual Property India has published a patent application (202641105222 A) filed by Dr. M. G. R. Educational & Research Institute on September 01, 2026, for Ai-Based Multilingual Employment Scam Risk Analysis And Offline Fraud Detection For Recruitment Platforms.
Inventors include Eswar G; Dwarakesh C R; V. Vaidehi; M. Gangadevi; and Viji Vinod.
The application for the patent was published on September 11, 2026, under issue no. 37/2026.
Abstract: Disclosed is a system (100) for analysing job postings, internship offers, and employment-related communications to detect employment-scam risk. The system (100) includes an input interface (102), a language-detection engine (104), a text-processing engine (106), a personally identifiable information anonymisation engine (108), and a feature-extraction engine (110). A model-routing engine (112) selects an English transformer model or a cross-lingual transformer model according to a detected language. A weighted ensemble prediction engine (114) combines outputs of a fine-tuned BERT model (116), an XLM-RoBERTa model (118), a TF-IDF and random- forest classifier (120), and a rule-based heuristic engine (122) through an aggregation engine (124) to generate a unified scam-risk score. A post-prediction override engine (126) modifies the score, while an explainability engine (128) generates line-specific multilingual explanations. A victim-report matching engine (130) identifies related reports. An output interface (132) presents the result. An edge-computing device (134) enables offline operation through a local wireless hotspot. Fig. 1 is the reference figure
Disclaimer: Curated by HT Syndication.
