Paper Key : IRJ************384
Author: Pilivkar Shubham,Wadekar Tushar,Khedkar Shrikant,Waghumbare Sharvari,Prof.ms.gaikwad S.
Date Published: 15 Apr 2024
Abstract
The problems of false information and information overload in the digital age are addressed in News Nexa: Beyond Headlines. It uses a variety of deep learning models to give users a comprehensive news experience.Taking On False News: News Nexa uses RNNs (LSTM, GRU), BERT, and T5 to classify news stories as authentic or fraudulent. This allows users to make judgments based on reliable information.Improved Content Arrangement: Users may easily browse through a plethora of online content thanks to the automatic classification of news articles into niche categories (sports, business, etc.) by deep learning models (RNNs, LSTM, GRU, BERT, T5).Better Information Consumption: GPT-2 models and transformers produce succinct and insightful summaries of news stories. This is very helpful for people who don't have a lot of time and can quickly understand the main points of a news article without reading it cover to cover.Breaking Down Language Barriers: News Nexa increases accessibility by utilizing Transformers and GPT-2 to translate news stories from English to Hindi. It may also include Marathi translation in the future by utilizing comparable models and open-source translation libraries.The goal of this research is to investigate the practical applications of several deep learning models for Natural Language Processing (NLP) problems. News Nexa has the power to completely transform the way people consume and engage with news, promoting a more knowledgeable and interconnected world community.Keywords: BERT, GPT-2, Transformers, RNNs, Deep Learning, News Summarization, News Classification, Fake News Classification, and Machine Translation Deep Learning, Transformers, RNNs, BERT, GPT-2, Mistral7B.