michaelchen-lab / SEAT

Sentiment analysis web app for non-technical users.

SEAT Web App

The SEAT Web App is a sentiment analysis web app which aims to democratize sentiment analysis for non-technical users.

Links

Article

• Final Report for SEAT

Video

• Youtube tutorial

Report

How It Works

This is a sentiment analysis tool meant to help users extract tweet data on their interested topic, and run sentiment analysis on that data.

Step 1: Extraction of tweets through Twitter API using keywords chosen by users

Step 2: Filter out irrelevant data and discard them using keywords chosen by users

Step 3: Categorise relevant data using topics chosen by users, and their corresponding keywords.

Step 4: Sentiment analysis is performed using pre-trained NLP model(s)

Get Started

Setup local Python environment

$ virtualenv venv

Enter Twitter consumer key and consumer secret key into env variables

Environment Variables: 'TWITTER_CONSUMER_KEY' and 'TWITTER_CONSUMER_SECRET'

Edit your PostgreSQL database credentials in /projects/settings.py

DATABASES = {
    'default': {
        'ENGINE': (engine),
        'NAME': '(name)',
        'USER': '(user)',
        'PASSWORD': '(password)',
        'HOST': '(host URL)',
        'PORT': '(port number)'
    }
}

Start Django local server in command line

$ python manage.py runserver

Go to the localhost URL provided by Django, and have fun exploring SEAT!