Deep-Sentiment-Analysis

Description

Sentiment analysis of twitter tweets using deep learning techniques

Models implemented so far

• LSTM model
• Fully connected neural network model which trains on doc vectors

Environment

Python2, Linux system (Preferable)

Memory requirements

• 4GB RAM recommended for LSTM model
• 8 GB RAM recommended for FCNN model

Dependencies

• python2
• pip (for python2)
• cuda library (if you want to use gpu accelaration for training)
• once you install these dependencies run pip2 install -r requirements.txt

Layout

• src/ folder which contains all codes
  • config.py contains the config of all the parameters needed
  • model.py contains the abstract class for all the models
  • test_results.py contains class to hold results of testing a model
  • <name>_model.py contains implementation code for corresponding models
  • utils.py contains utility functions
• data/ which contains training and test data
• models/ directory where the models are stored