Coffee Surf

• 

  Product page for one of the coffees.

What it does

• We built a search engine for finding a coffee that you already like or a coffee with certain characteristics
• We can then make recommendations based on the taste profile

How we built it

• We found a review database for coffees that had a poor search results and made no recommendations
• We first extracted features from the descriptions with NLTK
• We then used KNN to find the best recommendations for each coffee
• We started the front end with a template and we adapted it so that it shows the data from Flask
• We ended up implementing our own simple search engine

Challenges we ran into

• We had a lot of trouble starting out each step since we were not familiar with many tools we needed
• We especially had trouble making a font-end that works with the processing we did before since we both never really did web development.

Accomplishments that we're proud of

• It's our first project with NLP
• It's our first project that involves web-development

What we learned

• Fullstack development is hard!
• We learned how to use NLTK for the first steps of an NLP project
• We learned how to use templates in Flask to direct to a webpage based on Python backend

What's next for Coffee Surf

• We would first like to scrape data from more local roasters to be more useful
• Also, it would be really cool to generate a (pretentious) description from some keywords (think "almost a suggestion of vanilla"- James Hoffmann)

Built With

• css
• flask
• html
• natural-language-processing
• nltk
• python
• sklearn