MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641076800 A) filed by Vellore Institute Of Technology on June 21, 2026, for System And Method For Aspect-Level Sentiment Analysis Using Transformer-Based Sentiment Classification And Rule-Based Aspect Extraction.
Inventors include Baiju B V; and Bhavana Anand.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: ABSTRACT A system (100) and method for aspect-level sentiment analysis are disclosed. The system (100) comprises a data preprocessing module configured to generate cleaned review representations from review data, a sentiment classification module configured to process the cleaned review representations using a transformer-based language model and generate sentiment classifications and confidence scores, an aspect extraction module configured to identify aspects using predefined aspect-keyword mappings, and a sentiment mapping module configured to associate identified aspects with corresponding sentiment classifications. The system (100) further comprises a review prioritization module configured to rank reviews according to confidence scores, an explanation generation module configured to determine word-level contribution information for selected reviews, and an analytics generation module configured to aggregate sentiment classifications, aspect-sentiment associations, confidence scores, and explanation information. The disclosed architecture enables integrated aspect extraction, sentiment classification, confidence-based review prioritization, and explanation generation within a unified processing framework.
Disclaimer: Curated by HT Syndication.
