This is the source code of my submission for the Kaggle competition ["Bag of Words Meets Bags of Popcorn"] (https://www.kaggle.com/c/word2vec-nlp-tutorial). The leader board score is 0.9766.
The model is an ensemble of NBSVM, Paragraph Vector and Gated Recurrent Neural Network. It is based on the code from the paper ["Ensemble of Generative and Discriminative Techniques for Sentiment Analysis of Movie Reviews"] (http://arxiv.org/abs/1412.5335) by Grégoire Mesnil and others.
-
Data should be put into the directory data.
-
The code requires numpy, pandas, Theano and passage which can be installed by running:
sudo pip install -r requirements.txt
- After that, run the following command to generate the submission file "ensemble-submission.csv":
chmod +x run.sh && time ./run.sh
- The code was tested on Ubuntu 14.04. It took about 1h16' on an i7 quad-core with a GTX Titan. The submission should give the score of ~0.976 on the leader board (it can be slightly different due to data shuffling during training).
Since the code is based on Grégoire Mesnil's work at iclr15, all files except scripts/passage_nn.py are released under Creative Commons Attribution-NonCommercial 4.0 International License.