MUMBAI, India, Sept. 14 -- Intellectual Property India has published a patent application (202611086413 A) filed by Pranveer Singh Institute Of Technology,kanpur on July 15, 2026, for System And Context-Aware Method For Transcription Evaluation And Semantic Error Categorization.
Inventors include Mr. Ashish Tripathi; Dr. Piyush Bhushan Singh; Naman Chugh; Mr. Himanshu Kumar Diwedi; Dr. Vatsya Tiwari; Dr. Nayancy; Dr. Vivek Kumar; Dr. Sumit Chandra; Mr. Deepak Gupta; and Dr. Mohit Pandey.
The application for the patent was published on September 11, 2026, under issue no. 37/2026.
Abstract: The system uses a computerized approach together with its system and permanent computer-readable medium to assess human transcription skills through an asymmetric multi-document input system that grades typing performance. The system receives a source document and a target document which may independently comprise digital text files un structured data or optical images. In the event of optical image inputs, a decoupled optical character recognition (OCR) pipeline extracts alphanumeric and structural text strings ap plying dynamically assigned, asymmetric pre-processing parameters to normalize environ mental noise. A sophisticated sequence-matching algorithm computes primitive deviations between the extracted source and target strings. The contextual semantic filtering module executes two tasks by intercepting the primitive deviations which it uses to create different educational error categories through its character-class syntactical analysis method. The system establishes two distinct groups which include design layout errors that encompass whitespace and line breaks and punctuation errors together with actual alphanumeric vocab-ulary missing elements and replacement elements. The evaluation module applies predeter-mined fractional weight penalties mapped to each didactic category to calculate a composite performance accuracy score. The system further generates a synchronized graphical user in-terface mapping the categorized discrepancies via distinct visual markers onto a side-by-side rendering of the multi-modal inputs.
Disclaimer: Curated by HT Syndication.
