Sentiment Analysis is used to discover people’s opinions, emotions and feelings about a product or service. This model will try to predict the sentiment of the movie reviews with positive or negative.
The problem is to determine whether the given movie review has a positive or negative sentiment.
Dataset used in this problem is NAVER Review Dataset. It contains 200,000 movie reviews( positive or negative) for original, 150,000 movie reviews( positive or negative) for training and 50,000 movie reviews( positive or negative) for testing.
- Python 3.6
- Pytorch 1.2
- KoNLPy
- Mecab
- nltk
- matplotlib
- numpy
- wordcloud
According to result data, top accuracy has shown on Train Step 8. (87.59%)
On the model and summarizing the estimated performance. We can see that this model achieves an accuracy of 87.59%. Again, there is a lot of opportunity for further optimization, such as the use of deeper and/or larger layers.