This repository was created for my studies at IU International University for the course Project: NLP.
The preprocessing steps are located in the file preprocess.ipynb. The sentiment analysis with supervised machine learning models (BernoulliNB and SVM) are located in file supervised_NB_SVM.ipynb. The Deep Learning model with word embeddings is located in file deep_learning.ipynb. A small function that allows classification of user customed reviews is located in file predict.py.
In the folders, there are some plots of metrics, the resulting models saved (except the SVM model, out of size restrictions by github) and the logs of tensorboard.
For stable usage of the application, python version 3.10 is recommended. Install python from the official website. Check your python version with entering your command promt and execute the following command:
python --version It is recommended to use a customized environment to ensure full functionality, e.g. with the distribution anaconda, which can be downloaded here.
Install the required packages with the following command in your command line interface (For more information about pip, please check the pip documentation):
pip install -r requirements.txt
First download the code by either downloading the .zip-File or clone it via the command promt. For more information about the later please check the github docs.
To predict custom sentiment on movie reviews, either open the project with a python-IDE and run the file predict.py or enter the following in the command prompt:
python predict.py Of course it is also possible to run it in the Jupyter Notebook instances. Look there for the end of the files.