StockSense

Inspiration

We both occasionally dabble in the stock market, and noticed that researching stocks can take a long time, and thought that we could make a website do it for us.

What it does

It scrapes information from Yahoo Finance about articles related to the desired company, then detects the sentiment/opinion of the articles, and generates a summary of the current news related to the company.

How we built it

We used BeautifulSoup to scrape the data from Yahoo Finance, then used Language models such as vaderSentiment, spaCy, and Hugging Face Transformers to determine the sentiment of the text, then generate a summary of the finance related portions of all of the related articles.

Challenges we ran into

Because neither of us had used any of these before, we had to learn Natural Language Processing, Web Scraping, and even figure out how to build a website with Flask. Particularly, we had trouble navigating Yahoo's html tree, and figuring out the syntax regarding Flask reading the html.

Accomplishments that we're proud of

Given how little information we started off with, we managed to stick to our original plan and not lose any functionality, which is our biggest accomplishment and something we rarely manage to do.

What we learned

We had used almost none of these tools before, and we ended up learning not only about web scraping, but also using Language model api's, as well as creating rudimentary websites with flask.

What's next for StockSense

Our plans for StockSense are pretty simple, we would like to upgrade the web scraper to also cover more news sites than just Yahoo Finance, and we would also like to tweak the models we're using to hopefully handle more data from more articles.

Built With

• beautiful-soup
• flask
• html
• python
• requests
• spacy
• transformers
• vadersentiment