MUMBAI, India, July 24 -- Intellectual Property India has published a patent application (202611056499 A) filed by Baba Farid College Of Engineering And Technology on May 04, 2026, for Hardware-Based Token Filtering System And Method For Reducing Computational Load In Natural Language Processing.
Inventors include Dr. Pardeep Mittal; and Shubham Kumar.
The application for the patent was published on July 17, 2026, under issue no. 29/2026.
Abstract: The present disclosure relates to a hardware-based token filtering system (100) for reducing computational load in natural language processing. The system (100) comprises an embedded token filter unit (102) configured to perform parallel bit-pattern matching on an input character stream represented as a sequential bitstream to identify non-essential tokens. A semantic integrity preserver unit (104) evaluates whether removal of the identified tokens preserves semantic completeness and syntactic validity, and a bidirectional processing core unit (106) selectively suppresses the identified tokens to generate a filtered output stream. A computational load estimation unit (108) determines reduction in token count and processing effort. The system further includes a multilingual token dictionary (110), a memory unit (112), and an interface communication unit (114) for facilitating processing and data flow, thereby reducing computational load and energy consumption while maintaining contextual accuracy. FIG. 1 is the reference figure.
Disclaimer: Curated by HT Syndication.
