Sentence-Level Sentiment Analysis
One model to learn them all (both language model and sentiment analysis -> sentiment2vec)
- se-v1.ipynb: using sum all LSTM output vectors to predict sentiment label.
- se-v2.ipynb: using sum all LSTM output vectors of sentiment word to predict sentiment label.
- se-v3.ipynb: using sentiment embedding instead of one-hot vector.
- https://www.cs.york.ac.uk/semeval-2013/task2.html
- 67% accuracy rate
- tensorflow 1.2
- ipython notebook
Binh Do