Sentiment Analysis on Turkish Texts

In this project, I have made a simple web application to analyze sentiments on Turkish texts. I used an e-commerce dataset which had "Rating" and "Review" columns and contains roughly 250.000 commments. Rating column was tagged as 1 for positive comments, 0 for negative comments. In the model, 3-layered LSTM architecture was used. At the end of training, 95% accuracy was obtained. All codes were written in Python and Flask was used as a web server.

For details, please check the documentation

Instructions

git clone https://github.com/ozturkaslii/analyze-turkish-sentiment.git
cd analyze-turkish-sentiment
pip install -r requirements.txt