Inspiration

To help find relevant global news automatically sourced to be easily read and understood. There is a lack of global news that does not contain bias towards an opinion or certain region. We wanted to aggregate news into headlines that are easy to understand and quick to sort through as well as provide a clear indication of the reliance of the news.

What it does

The Global News Map pulls news headlines from reputable news sources (e.g Reuters news wire), which is processed using our proprietary language processing model to gauge its geopolitical and global relevance and impact. Headlines deemed globally relevant are labelled with the countries that they are associated with and can be accessed on a map vizualization - you can know what's happening in a specific part of the world in a single click.

How we built it

We built the front end using React, specially the J-Vector Map package for the global map visualization. The backend is written in python using Flask. The language processing model is also built on python using the nltk, geotext and textblob libraries. We're using a MongoDB database.

Challenges we ran into

  1. Extracting country associations in the language processing model and building a metric to gauge the objectivity of the headline.
  2. Integrating the services with each other.
  3. The distributed nature of collaboration resulted in minor complications with package management

Accomplishments that we're proud of

Getting a fully functional proof of concept running. Accurately gauging objectivity to an appreciable measure by developing a novel metric Integrating information streams in real-time

What we learned

NLP is hard, more for epistemological reasons than computational reasons. We could have benefited by investing more time in the initial structuring.

What's next for Global News Map

Integrating multiple news sources and improving the language processing model even further.

Share this project:

Updates