Sentiment Analysis

1. Data sets

• Twitter US Airline Sentiment
• Large Movie Review Dataset
• Multi-Domain Sentiment Dataset

1. Project structure

The project is based on several modules, each of which is responsible for one task related to the sentiment analysis.

Module loader

The module is responsible for loading data sets

Module preprocessing

The package contains methods for text preprocessing. For example:

• Punctuations
• Emoticons
• Negation
• Stemming
• Lemmatization

Module vectorization

The module contains tokenization methods

• Bag of Word
• Tf-Idf
• Doc2Vec

Module classification

Helper for the classification task based on GridSearchCV.

Module plot

This module defines several basic charts that help show the structure of the data.

Module experiment

This is the heart of the application (Main). We can find classes responsible for loading data and performing text preprocessing here. For data classification here