https://github.com/standardProton/hackcu10/
Inspiration
There are 624 total scholarships that can still be submitted, but reading the description of every single one would take forever. We need a search tool to improve this.
What it does
By pasting your resume, major, etc., the site can search the descriptions of all of the scholarships to find the most related ones.
How we built it
React, Python, JavaScript
Challenges we ran into
Scraping the data, size of the document vectors, caching the scholarship data, homepage design css.
Accomplishments that we're proud of
The site's speed is amazing after we redid the caching, and we were able to successfully scrape scholarship data in a usable format.
What we learned
- How to parse HTML in Python
- How to create an interval that makes the site animated (+advanced css topics)
- Natural Language processing and tokenization both in Python and JavaScript
- Cosine distance document similarity
- Python BeautifulSoup library
What's next for CU Scholarships
Using AI to match scholarships, such as an LSTM to create an embedding vector for each scholarship, which can be compared to an embedding vector from the user's input.
Log in or sign up for Devpost to join the conversation.