Implementation of Machine Learning and Deep Learning for Sentiment Analysis

• Sentiment analysis is the process of determining whether a piece of writing is positive, negative or neutral.
• In this project I have demondtrated how various Machine Learning and Deep Learning models can be used for sentiment analysis.

The Dataset:

• The dataset used is "Sentiment Labelled Sentences Dataset", from the UC Irvine Machine Learning Repository.

• The sentences come from three different websites/fields:

  • amazon.com
  • imdb.com
  • yelp.com

• Each sentence is labelled as either 1 (for positive) or 0 (for negative).

• For each website,tThere exist 500 positive and 500 negative sentences.

• This dataset was created for the Paper 'From Group to Individual Labels using Deep Features', Kotzias et. al,. KDD 2015. (Please cite the paper if you want to use it :))

• Link to the dataset is: Sentiment Labelled Sentences Data Set

• The dataset is present in the Dataset folder.

Machine Learning models:

• I have used the follwoing Machine Learning models:

1. Multinomial Naive bayes
2. Random Forest
3. LinearSVC

• The code implementing these models is in 'modules/Sentiment_Analysis_ML.ipynb'.
• All the trained models are stored at 'models/ML'. Thereafter the models are segrated as per the dataset (Amazon, IMDB, Yelp).

Deep Learning models:

• I have used the follwoing Deep Learning models:

1. Feed Forward Neural Network (FFNN)
2. Convolutional Neural Network (CNN)
3. Recurrent Neural Network (LSTM)

• As the dataset consists of three different set of data, I have created three different implementations for each of them.

1. Amazon product Rreview Dataset ('modules/Amazon_Sentiment_Analysis_DL.ipynb')
2. IMDB Movie Review Dataset ('modules/IMDB_Sentiment_Analysis_DL.ipynb')
3. Yelp Restuarant Review Dataset ('modules/Yelp_Sentiment_Analysis_DL.ipynb')

• All the trained models are stored at 'models/DL'. Thereafter the models are segrated as per the dataset (Amazon, IMDB, Yelp).

Word Embeddings:

• All the Deep Learning architectures use the GloVe Word Embeddings.
• To download click here (please download them before running the code.)
• The 6 Billion words, 100 dimensional vector representation variant is used.
• The have been stored at location 'Dataset/GloVe_Word_Embeddings'

Results:

After tyring various machine learning and deep learning models, I got the following results.

Model	Amazon Reviews	IMDB Reviews	Yelp Reviews
Multinomial Naive Bayes	85%	85%	78%
Random Forest	80%	79%	79%
Linear SVC	84%	81.50%	80%
FFNN	81.50%	84%	82%
CNN	87%	85.50%	82.50%
LSTM	87%	85%	83%