MUMBAI, India, July 24 -- Intellectual Property India has published a patent application (202611062465 A) filed by Gaurav Srivastav; Richa Singh; Rajeev Kumar Singh; Bhawna; Piyush Agarwal; Shivansh Prasad; Abhishek Shukla; and Anuraag Raj Narayan on May 17, 2026, for Edbert: A Unified Transformer-Based Framework For Cognitive And Affective Assessment In Education.
Inventors include Gaurav Srivastav; Richa Singh; Rajeev Kumar Singh; Bhawna; Piyush Agarwal; Shivansh Prasad; Abhishek Shukla; and Anuraag Raj Narayan.
The application for the patent was published on July 17, 2026, under issue no. 29/2026.
Abstract: A transformer-based system for automatic analysis of student feedback in educational contexts is disclosed by the current invention, G7SortED. The innovation processes student-generated text reviews and extracts valuable insights using a refined BERT (Bidirectional Encoder Representations from Transformers) model. The system has modules for explainability, output generation, dual-layer classification, contextual embedding, and data preprocessing. According to Bloom's taxonomy of educational aims, the categorization module is set up to concurrently identify affective and cognitive variables, including stress, motivation, engagement, and satisfaction. The explainability module ensures transparency for educators and institutions by offering interpretable insights through attention-based visualizations and SHAP-based feature attribution. Additionally, the invention includes assessment and deployment modules that facilitate smooth interaction with institutional dashboards and benchmarking against machine learning baselines. The invention overcomes the drawbacks of conventional sentiment analysis and manual review processes by fusing explainable AI with domain-adapted natural language processing techniques. This results in a framework for improving educational quality that is accurate, scalable, and interpretable.
Disclaimer: Curated by HT Syndication.
