Automated Multilingual Crawling, Categorization, and Sentiment Analysis of Government-Related Digital News with Real-Time Feedback Integration

View at publisher Summarise with VIRAT
Authors: Aparna R. Sawant, Rishi Raj Jain, Rugved Junghare, Sai Kadam, Ishita Jasuja, Vinayak Wankhade
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

With the advent of today’s digital era, news sources provide different views, and therefore it is essential to analyze and classify news so that an opinion can be made based on the same. There has been a system developed for processing news content in English, Hindi, and Marathi, collecting more than 13,000 articles and videos from 15+ credible sources, including government and public news sources. This system ranks news stories into ten preselected categories according to government departments, making use of machine learning classifiers learned from 9,000 tagged samples. The sentiment analysis, which is driven by a multilingual BERT-based model, classifies the tone of every article as positive, neutral, or negative. This model has seen an 90% accuracy in sentiment prediction. About 18% of the articles in the data set are labeled as negative and trigger an automatic email alert to the respective government department for immediate action. The system has a responsive, web-based interface where users can see categorized news cards with language, sentiment, and department filters. The cards provide summaries, sentiment labels, and links to full articles. Users can dynamically update or sort sentiment scores, and the system has real-time filtering for all three languages. Future plans involve increasing support for languages in five additional regional languages, model accuracy to surpass 90% using larger and more diverse sets of data, sentiment analysis with more in-depth contextual understanding, and mobile applications for Android and iOS. The platform is constructed on a cloud-based infrastructure with the ability to support thousands of simultaneous users for fast response times and real-time updates. The purpose of the system is to promote fair news consumption and equip citizens with reliable information, ultimately aiding in transparent governance and social harmony. Through enabling policy decisions based on data, it aims to aid in an educated public and better quality of life.

Fields of Study & Subjects

Sentiment Analysis and Opinion Mining Text and Document Classification Technologies Web Data Mining and Analysis

Publication details

Journal / source
CRC Press eBooks
Pages
147–155
Publication date
26 March 2026
Language
EN
Document type
Book chapter