GloVe-CNN-BiLSTM Model for Sentiment Analysis on Text Reviews

Nowadays, social media networks generate a tremendous amount of information from their users. To understand people’s views and sentimental tendencies on commodity or an event timely, it is necessary to conduct text sentiment analysis the expressed by For microblog comment data, always mixed with long short texts, which relatively complex. Especially for contains lot content, correlation between words more complex than that in text. study classification these texts composed long-text short-text, this research proposes optimized GloVe-CNN-BiLSTM-based model. In model, GloVe used vectorize words, CNN given represent part space character. BiLSTM build temporal relationship. Twitter’s data COVID-19 as experimental dataset. The results experiments suggest method can effectually identify tendency users’ online comments, accuracy complete-text, long-text, short-text achieve 0.9565, 0.9509, 0.9560, respectively, obviously higher other deep learning models. At same time, show has good field expansion.