This repository contains scripts for:
- Collecting tweets from Twitter Streaming API
- Performing sentiment analysis on collected data
- Training Convolutional Neural Network for emotion classification
- Using trained CNN model for emotion classification
Python >= 3.5
virtualenv
In order to use script for tweets collection, copy collect/config_example.ini
file to collect/config.ini
and fill it with your Twitter application credentials.
Install required dependencies:
pip install -r requirements.txt
Anaconda or Miniconda is recommended for running this project on Windows.
Create new Conda environment and install Conda dependencies:
conda install --file conda_requirements.txt
And install pip dependecies with:
pip install -r requirements.txt
Dataset used for training models is based on Sentiment Analysis: Emotion in Text dataset and can be downloaded from this location: https://s3.eu-central-1.amazonaws.com/tea-datasets/text_emotion_augmented_v2.csv