Variants of Long Short-Term Memory for Sentiment Analysis on Vietnamese Students' Feedback Corpus

The Long Short-Term Memory (LSTM) and Dependency Tree-LSTM have shown the state-of-the-art results for the sentiment analysis task for the English language. Despite many studies of LSTM approach, there are no studies of Dependency Tree-LSTM approach for Vietnamese sentiment analysis. In this paper, we conducted experiments with LSTM, Dependency Tree-LSTM, and our proposed models on Vietnamese Students' Feedback Corpus. According to the experimental results, the Dependency Tree-LSTM were not better than the LSTM model. However, when combining final hidden state vectors of LSTM and Dependency Tree-LSTM models with a Support Vector Machine classifier, we achieved the F1-score of 90.2%, which is higher than the performance of the LSTM model.