Authors : Aymen Samir, Saleh Mesbah and Magda Madbouly
Abstract: Bidirectional Encoder Representations from Transformers (BERT) represents the latest technology of pre-trained language models which have recently advanced a wide range of natural language processing tasks. This study aims to investigate how BERT can be usefully applied in sentiment analysis tasks with fully connected neural network. The proposed model is developed using simple tips preventing it from over-fitting and enabling it to be fine-tuned easily on such down stream task. BERT performs much better as a Strong text embedding Model. Using such procedures successfully provide a better accuracy than the expensive machine learning procedures.
Aymen Samir, Saleh Mesbah and Magda Madbouly, 2021. Artificial Intelligence-Based Sentiment Analysis. Journal of Engineering and Applied Sciences, 16: 18-22.