Movie Review Sentiment Classification with Naive Bayes Classifiers
This is a mini project implemented in April 2020 for Introduction to Information Retrieval course in Bogazici University.
3 types of Naive Bayes classifiers (Multinomial, Bernoulli, and Binary) were implemented to carry out sentiment classification (positive/negative) on Cornell Movie Review Dataset (polarity v2.0).
To run the code, please put the naive_bayes.py file in a directory which contains the data folder. Dataset must be unzipped. Here is an example directory:
Working dir
|
|------ naive_bayes.py
|
|------ data
|
|------ train
| |
| |------ pos
| |
| |------ neg
|
|------ test
|
|------ pos
|
|------ neg
I've also attached auxilary .npy and .txt files to reduce run time. They should be put in the same directory as naive_bayes.py. Code works without those auxilary files but takes considerably longer time. Code doesn't take any command line arguments.