Inspiration

Many times, I don't have the time or patience to read through all documents on the internet when researching companies and their financials with the purpose of investing. This tool drastically reduces the amount of time taken to scavenge the internet for reliable articles and read each one.

What it does

This tool scrapes the web for reliable articles and summarizes them several times until resulting in a final paragraph.

How we built it

Web scraper

I built a web crawler that effectively scrapes the web for articles and parses them

Article filter

I built a filter that effectively filters the articles by reputation of source

Summarization pipeline

I built an article summarizing pipeline that effectively summarizes articles given important parameters

Challenges we ran into

I ran into getting my Python code working with Poe as there were many type safety and casting errors that came up when deploying my backend which were not present on my machine. Thus, I was not able to introduce my summarizing technology pipeline (3) in my demo today. However, I will show my backend in action.

Accomplishments that we're proud of

I am proud that I created a tool that is useful and will save me a lot of time while also being able to scale to meet an everlasting set of custom parameters. I can see myself working on this project for a long time to provide increasingly better output.

What we learned

I learned that I should not underestimate the amount of time it takes to deploy :p. On a more serious note, I learned that there is are so many new platforms out there to publish my technology onto. I really like Poe because it is very developer-friendly and is quick to get up and running. I will continue to use Poe after this hackathon and want to finish this project and make it even better.

What's next for Findown: Aggregate & Summarize All Financial News

  • I will add more inputs parameters (more financial data, more news sources)
  • I will add more outputs (overall sentiment, metrical output - "Rate stock on a scale of 1-10")

Built With

Share this project:

Updates