The Role of ERNIE Model in Analyzing Hotel Reviews using Chinese Sentiment Analysis

This research aims to find the best deep learning model to do Chinese sentimental analysis. BERT’s model may work well in the English language but not work in the Chinese language. English is easier to encode in embedding space, and words are separated by space characters, unlike Chinese words, which do not exist separate by the area. BERT Chinese model work on character level encoding is not enough to make the BERT model adapt to Chinese NLP tasks. The BERT Model and ERNIE models are the current state of art pre-training language models. Word level embeddings and input is needed to improve the BERT model to perform in Chinese Sentiment Analysis. With the motivation to improve the Chinese Sentiment Analysis, this study will explore the role of different models to propose a better version of the ERNIE model. This study will limit the scope to improve Chinese sentiment analysis among different NLP tasks.