NextGlass

working on redeploying

Background and Functionality | Technologies

NextGlass is a web application that uses your preferences in wine to suggest new wines to taste.

Background and Overview

Make finding drinks easier. Rather than asking around, get recommendations from what you like. Let's try to find your NextGlass!

Functionality and MVP

• User auth: sign up, login, logout
• Collection: create, edit, delete
• Collection page
• Landing page with categories
• Production README
• Team Links

Bonus Features

• Search
• Future: suggestions with tensorflow

Technologies and Technical Challenges

Backend: MongoDB and Express

• Relevant data is stored in columns
  • Some of the columns contains an array, containing some of 50+ tags
• Tables: users, collections, wines

Technical Challenges:

• Fetching data efficiently. We leveraged mongodb search index to enable fuzzy searching.
• Cleaning data. We used excel to:
  • Remove all incomplete entries
  • Create a list of common tags from the content of the descriptions
  • Remove entries without the most common tags
  • Trim down the 130k dataset to 10k
• Manipulating data
  • Original data on kaggle
  • split.js contains the final code we used to change the tags and tagindex data from strings to arrays
  • symbols.js contains the final code we used to update the description data to use UTF-8 characters

Frontend: React, Redux, Axios

• Landing page contains images linked to curated collections

Technical Challenges:

• Double fetching- fetching a collection, and then dependent wines. Final solution using thenable promises, as componentDidUpdate will be deprecated.
• Creating a recommendation model for suggesting drinks based on tags from the current collection
• Reading data from the MongoDB database to feed into the recommendation model

Group Members and Work Breakdown

Katherine Wu, Malachi Coberley, David Chan

Day 1

• Discuss database structure, querying - All
• Discuss UI/UX, sketch out wireframe - All
• Create routes, configuration, databases, possibly begin user auth - David
• Build sign up/login view - Malachi
• Begin creating recommendation model - Katherine

Day 2

• Finish up user auth, start collections - David
• Simple recommendation model - Katherine
• Landing page skeleton, start individual collection page skeleton - Malachi
• Discuss how to implement curated collections - All

Day 3

• Wine index component - Malachi
• Change wine tags data type - Katherine, David

Day 4

• Wine component - David
• User's collections component - Malachi
• Tags query and component - Katherine

Day 5

• Suggestions component - Malachi
• Collection's wines component - David
• Mongoose query and index for fuzzy search - Katherine

Day 6

• Complete production README - Katherine
• Search component, about and credits components - Malachi
• Delete wines; refactor collections - David