Sentiment-Analyzer

Also a sarcasm detector now

Different models to train

• IMDB movie review based sentiment analysis (either positive or negative)
  • Run with option flag '--type imdb'
• Twitter Sentiment140 based sentiment analysis (either positive or negative)
  • Run with option flag '--type twitter'
• Twitter sarcasm detection (either sarcastic or not)
  • Run with option flag '--type sarcasm'

Ways to run repl

• Pure sarcasm detection
  • Run with option flag '--use_sarcasm'
• Sentiment analysis without sarcasm detection
  • Specify dataset to be run with using '--type [imdb or twitter]'
• Sentiment analysis with sarcasm detection (If sarcasm is detected, sentiment is flipped)
  • Run with option flag '--use_sarcasm' and '--type [imdb or twitter]'

Datasets

• Elaborate more on these Below are the datasets used IMDB Sentiment140, 1.6 million tweets each tagged with a sentiment 0-4 Sarcasm on Reddit - https://www.kaggle.com/danofer/sarcasm?select=test-balanced.csv Just download train-balanced-sarcasm.csv

Setup

How to run