GitHub - chymchym1905/VN-apple-watch-sentiment-analysis: Sentiment analysis using CNN, RNN and CRNN with custom PyTorch model

Check out the report

Experiment between CNN, RNN and CRNN on a non-optimal, high context dataset. Highest accuracy achieved is only 64% even though I did a lot on dataset optimization.

Name		Name	Last commit message	Last commit date
Latest commit History 10 Commits
.vscode		.vscode
res10-8.8k-simplelemma-1layer-CRNN		res10-8.8k-simplelemma-1layer-CRNN
res16-9k-3layers-RNN-nocleantext		res16-9k-3layers-RNN-nocleantext
res19-8.8k-simplelemma-1layer-CNN		res19-8.8k-simplelemma-1layer-CNN
.gitignore		.gitignore
CNN+CRNN.ipynb		CNN+CRNN.ipynb
CNNtest.py		CNNtest.py
README.md		README.md
RNN.ipynb		RNN.ipynb
cleantext.png		cleantext.png
datadist.png		datadist.png
report.pdf		report.pdf
vector.png		vector.png
vi-model-CBOW.bin		vi-model-CBOW.bin
vietnamese-stopwords.txt		vietnamese-stopwords.txt
vlsp_sentiment_test - Copy.csv		vlsp_sentiment_test - Copy.csv
vlsp_sentiment_test.csv		vlsp_sentiment_test.csv
vlsp_sentiment_train - Copy.csv		vlsp_sentiment_train - Copy.csv
vlsp_sentiment_train.csv		vlsp_sentiment_train.csv
word2vec+[crnn, cnn,rnn].ipynb		word2vec+[crnn, cnn,rnn].ipynb