Abstract
In the era of rapid information growth, efficiently processing news content in multiple languages has become increasingly important. This project presents a hybrid News Classification and Summarization System capable of handling both English and Kannada news articles. The system leverages natural language processing (NLP) techniques, including text cleaning, translation, TF-IDF vectorization, and machine learning classification, alongside transformer-based models (Pegasus) for abstractive summarization. Kannada text is first translated to English for improved classification accuracy and then translated back for user comprehension. Additionally, a keyword-based flood detection module highlights urgent news content. The system provides a unified platform for news categorization, summarization, and bilingual support, enabling users to quickly grasp critical information while ensuring accessibility across languages. Experimental evaluation demonstrates satisfactory performance in terms of accuracy and summarization quality, making it suitable for real-time news processing applications.
Keywords:
News Classification, Text Summarization, Natural Language Processing (NLP), Pegasus Transformer, Kannada-English Translation, Machine Learning, TF-IDF Vectorization, Bilingual News Processing.









Introduction
With the exponential growth of digital content, news articles are being generated in multiple languages at an unprecedented rate. The ability to efficiently process and extract meaningful information from these articles is crucial for users, organizations, and government agencies alike. Traditional manual methods of news analysis are time-consuming, error-prone, and unable to scale with the volume of data generated daily.
This project addresses these challenges by developing a bilingual News Classification and Summarization System that supports both English and Kannada languages. The system combines machine learning techniques for news categorization with state-of-the-art transformer models for abstractive summarization, providing a comprehensive solution for information extraction. Kannada text is translated to English to leverage the higher accuracy of existing English NLP models, after which summaries and categories are translated back to Kannada to ensure accessibility for regional language users.
Additionally, the system integrates a keyword-based alert mechanism for critical events such as floods, allowing urgent news to be identified and highlighted promptly. By automating news classification, summarization, and bilingual support, this project enables users to quickly comprehend large volumes of news content while maintaining linguistic inclusivity. The implementation demonstrates the synergy of translation, machine learning, and deep learning models in creating an efficient and scalable news processing platform.

Objective of the project
The primary objective of this project is to develop a bilingual News Classification and Summarization System that efficiently processes news content in English and Kannada. The specific objectives are as follows:
1. Automated News Classification:
o Classify news articles into predefined categories using machine learning techniques for both English and Kannada texts.
2. Text Summarization:
o Generate concise, meaningful summaries of news articles using transformer-based models, enabling users to grasp key information quickly.
3. Bilingual Support:
o Implement accurate Kannada-to-English and English-to-Kannada translation, allowing seamless processing and accessibility across languages.
4. Critical Event Detection:
o Identify urgent or disaster-related news (e.g., floods) using keyword-based and machine learning approaches for timely alerts.
5. Integration of NLP Techniques:
o Combine TF-IDF vectorization, machine learning classifiers, text cleaning, translation, and transformer-based summarization in a single workflow for efficient news processing.
6. User-Friendly Platform:
o Develop a web interface that allows users to input news text, select language, and receive categorized summaries with minimal effort.
7. Scalability and Real-Time Processing:
o Ensure that the system can handle large volumes of news articles efficiently, providing real-time categorization and summarization.