Sentiment Analysis using TensorFlow

This is a basic neural network in TensorFlow with one input layer and two hidden layers to identify sentiment in sentences.

The dataset used belongs to a Stanford project called Sentiment140 and can be downloaded here.

Also this neural network is using basic techniques during the data preparation phase, feel free to improve that.

How to run the code

After downloading the repo, open the 'use_neural_net.py' file and uncomment line 46 so your neural network can be trained. Once trained use the 'use_neural_network' function providing a sentence as parameter to get its sentiment.

Files description

• data_prep.py - contains all the functions responsible for preparing all the sample sentences and let them ready to be fed into the neural network.
• sentiment_neural_net.py - where the neural netowrk model is created, all the training happens and the model is saved.
• use_neural_net.py - kick off code used to call the training function as well as use the neural network.