The goal of this project is to enable a user to build an audience profile of Yelp users who have written reviews on the service, based off of a specific category of restaurant.
A user can search for businesses by category (i.e. 'mexican', 'cantonese') or search for all users who have written reviews for the same list of businesses.
Note: This is a fork of a previous coding assignment meant to test whether a database is needed at all for this project. I am planning to proceed without one as of writing (8/31/20)
To start the docker containers you may run make run or docker-compose build and then docker-compose up.
You may use any of the following end points to query the data:
/<category>/search/search/<category- the main search function of the application, this allows you to obtain a list of businesses and users that match this category/categories- view a list of all available categories/business/<category>- view a list of all businesses near a specified address that contain a specified category/users/<category>- gather a list of users that have written reviews for restaurants with this category
Occasionally, docker inspect yabdb_docker is needed to fetch the IP address for the container, which is used by the backend app.py to connect
to the db. At a later date this could be fixed.
Currently, the application will call the Yelp API's /businesses/search endpoint with a category and address parameter in order to obtain, and then save, a list of businesses that possess the category and are near the specified address. The data gets saved into a PostgreSQL database table called businesses and the categories are mapped to the businesses in a business_categories table.
To find the user list, the Yelp /businesses/{id}/reviews endpoint is called, which returns a list of three reviews sorted by Yelp's default sorting algorithm. Each review has data on the user that wrote the review, so a user object is created and saved to the database in a users table. Right now a user can only exist for one review.
More work could be done to build out the user data into a more valid form of audience profiling. Yelp does not offer much in the way of user querying but round-about ways of identifying user data with each user's unique ID could be done. Web scraping of a user's reviews on their profile page could reveal more information on their preferences - perhaps how frequently they visit a category.
The project has met the needs of the assignment as follows:
- A backend service was created with Python
- Calls a third party API - Yelp
- Collects two pieces of connected information - Businesses & Users that have written reviews, Categories & Businesses that possess those Categories
- Stored in a PostgreSQL database with tables: categories, businesses, users, business_categories
- Run in a docker container using docker-compose
- Create a README