Sentiment Analysis

The task is to identify sentiment analysis of amazon reviews and twitter tweets.

Amazon reviews: Implementation of Naive Bayes classifier after pre-processing the reviews to clean and label them.

Twitter Tweets: Implementation of Logistic Regression, Naive Bayes and Decision Trees classifiers after pre-processing the tweets. Also, used GloVe for word embeddings and ELMO to fetch contextual embeddings.

Dataset

I used open source data set of amazon reviews for this purpose.

Link: https://www.cs.jhu.edu/mdredze/datasets/sentiment/unprocessed.tar.gz

For tweets, scraped a small dataset of tweets off the website.

Requirements

1. Libraries required for running the python script:

• sklearn
• pandas
• numpy
• matplotlib

Run the script

To execute Amazon reviews sentiment analysis - python naive_bayes_reviews.py

To execute Tweets sentiment analysis- python log_res.py, python log_res_word_embed.py and python better_models.py

General notes

1. Correct the data folder location to where the reviews are present.

2. To test a particular review, use this line after executing the complete script: classifier.classify(get_features('returned tshirt .', all_words, 500))

Contact

Please contact [email protected] in case of any queries.