EnviroNLP

Inspiration

The inspiration behind this project was to create a topic modelling and scoring system capable of segregating companies and entities into different tiers of social, environmental and governance responsibilities

What it does

This project scrapes influential articles from any particular date and then scrapes the text from the article to run a topic modelling algorithm on it using expert.ai's NLAPI to create a visual representation of the different important and prevalent topics related to ESG.

How we built it

We used GDELT package to track important articles and then beautifulsoup to scrape the text from those same articles. NL-API from expert.ai was used afterward to help in making sense of the text with regards to ESG metrics.

Challenges we ran into

Expert.ai's API was sometimes unable to process larger amounts of data/text. Data cleaning was also one topic where we wanted to do a more comprehensive job, as the news articles when scraped gains a lot more noise than general, clean paragraphs.

Accomplishments that we're proud of

We explored the APIs offered by expert.ai for this project and were impressed by the capabilities it offers. We were also able to integrate GDELT in this project and if we are able to generate a robust scoring system, we hope to contribute to the fantastic work being done in GDELT in the open-source library space.

What's next for EnviroNLP

Our next steps would be to think deeply about how we can create a scoring system for this ESG data being scraped from the articles and integrate it with the GDELT open source dataset.

Built With

• beautiful-soup
• expertai-nlapi
• gdelt
• python