Here we used the dataset provided in the kaggle challenge link: https://www.kaggle.com/t/38fb9af3b5ab478db84d5cd8272265be
The dataset consisted of reviews of a certain place on Yelp! and labels associated with those reviews (could be interpreted as stars). Label meaning: 0 - negative review 1 - neutral review 2 - positive review
Our job was to classify a test set of reviews and assign them labels based on their content.
We have used various approaches to go about this problem and have found that LSTM, a modified RNN approach works the best among the ones chosen. We recorded a final accuracy of 90.727% on the test dataset.
This project was given to us as a condensed part of the course CS5242, National University of Singapore, as a 3 week summer internship program. (GAIP Big Data Analysis using ANNs)
- github.com/ishaanverma
- github.com/AnKuR-GaRg1
- github.com/mukeshkaranth
- github.com/janimalaga
- Samksha Bhardwaj