Predict the score of a wine review based purely on the textual features on the review.
Use Word Embeddings and LSTM to extract as much semantic value from the reviews. The hypothesis is that semantically positive reviews will have higher ratings that semantically negative reviews.
Below table contains files of relevance.
| File | Description |
|---|---|
EDA.ipynb |
Notebook containing the all code and process explanation. |
The following pages were used as inspiration for this project:
O'Reilly - Perform sentiment analysis with LSTMs, using Tensorflow