twitter-geo-sentiment

twitter-geo-sentiment is web app built with a ReactJS interface connected to a RESTful Flask API to analyse the sentiment of collected geo-tagged tweets, providing data visualisation based upon the collated data.

The Premium 30-day Twitter Search API is used to collect the most recent tweets based on given keywords by location. A naive-bayes sentiment classifier trained on tweets determines the sentiment of individual tweets, which are then aggregated to provide a mean sentiment for a region. A React frontend receives all this and provides data visualisations illustrating sentiment regarding a specific topic across states or countries.

This repository is the source code for my submission for the IB Computer Science Internal Assessment.

Installing

Prerequisites

• Twitter API (sign up and complete approval process – may take a few days)
• MongoDB Atlas (create account, create a cluster and connect to database)
• NodeJS

Dependencies

The JS dependencies of the ReactJS frontend are detailed in package.json.

The Python dependencies of the backend API are detailed in requirements.txt.

User installation

Clone the repository in a directory of your choosing:

$ git clone https://github.com/jasonfyw/twitter-geo-sentiment.git

Change directory into the cloned repository:

$ cd twitter-geo-sentiment

Then, install the JS dependencies from package.json using npm:

$ npm install

Now, change into the api directory to set up and enter a virtual Python environment using venv:

$ cd api
$ python3 -m venv venv
$ source venv/bin/activate

We can now safely install the Python dependencies and run the scripts properly. Using requirements.txt, we can do this pretty painlessly (although it might take a few minutes):

$ pip install -r ../requirements.txt

User setup

In the api directory, create two files with the following names and content contained within them.

.env – this file contains details to access your MongoDB cluster. Paste the following into the file. Then, replace [your_URI] with your URI connection string from MongoDB Atlas and [database_name] with the name you want to call the database (note: in your URI, make sure to include your password, as well as the value for [database_name]. To avoid an SSL certificate error, add &ssl=true&ssl_cert_reqs=CERT_NONE to the end of your URI as well):

MONGODB_URI=[your_URI]
DB_NAME=[database_name]

twitter_cred.yaml – this file contains your secret keys for access to the Twitter API. Paste the following, replacing [consumer_key] and [consumer_secret_key] with the respective values from your Twitter developer account.

search_tweets_30_day_dev:
  account_type: premium
  endpoint: https://api.twitter.com/1.1/tweets/search/30day/dev.json
  consumer_key: [consumer_key]
  consumer_secret: [consumer_secret_key]

Usage

You're now able to launch the web app! In the command line, start the backend API on your machine by running:

$ yarn start-api

Then, open another terminal prompt and navigate to the repository. Then, launch the ReactJS frontend by running:

$ npm start

You should now be able to access the web app at http://localhost:3000. It is also possible to access it from across your local network, however, adjustments need to be made to the API.

Contributing

As this represents the codebase for my IB Computer Science Internal Assessment, I will not be accepting contributions to this project. That said, feel free to fork this repo and go crazy.