NewsFriend Introduction
In today's day and age, people spam-share news to all their contacts often without reading the entire news article or without veryfiying if the article is the best version of the news beign shared.
The NewsFriend app attempts to address that by to pre-filtering a news article before reading it. To use the app, a user simply pastes a url into the app search bar, or shares the url via the phone's share button. The app will find then analyze the news article shared and find the best alternative articles for the given news. Furthermore, the app will rank your article based on readability/complexity of language, reading time/ article length, and the article's sentiment. After doing this analysis on all the alternative articles, the algorithm will give your orignal article a rating and suggest the better alternative articles on the same news for you to read! Here are the techinues utilized in this project
- Flutter for mobile app development
- Heroku + FastAPI to build a custom API to process the news articles
- WebScrapping to crawl google and figure out the most relavant alternative news sources for a particular news article
- WebScrapping for finding relevant images, the news company's name, and the new's text to use for readability, sentiment, and reading time metrics
- Sentiment and News categorizers built on Tensorflow/Keras using Word Vectors, LSTM based RNNs and other NLP techniques
Screenshots:
Check out the full code by clicking the picture below:
