MUMBAI, India, July 13 -- Intellectual Property India has published a patent application (202631081373 A) filed by National Institute Of Technology Jamshedpur on July 01, 2026, for A System And Method For Language-Aware Graph-Based Sentiment Classification Of Code-Mixed Text.
Inventors include Sahoo, Chinmayee; and Singh, Dr. Binod Kumar.
The application for the patent was published on July 10, 2026, under issue no. 28/2026.
Abstract: A SYSTEM AND METHOD FOR LANGUAGE-AWARE GRAPH-BASED SENTIMENT CLASSIFICATION OF CODE-MIXED TEXT The present invention discloses a computer-implemented system (100) and method for language-aware sentiment classification of code-mixed text. A code-mixed text input interface (102) receives a code-mixed text input (200), and a tokenization module (106) generates a token sequence. A language identification module (108) assigns language identifiers, and a token-language mapping module (110) maps word-level language labels to subword tokens. A multilingual contextual embedding generation module (112) generates a multilingual contextual embedding sequence (210). A language-aware edge-type matrix construction module (114) constructs a language-aware edge-type matrix (166) representing self-loop, adjacent-token, same-language local, and language-switch adjacent relationships. A language-aware graph attention module (116) refines the embeddings using edge-type-conditioned graph attention to generate a refined token representation (184). A sentiment classification module (120) classifies the input into a plurality of sentiment classes and a sentiment output interface (122) outputs a sentiment output (216).
Disclaimer: Curated by HT Syndication.
