Abstract
This study presents a technique for analysing customer analyses sentiment using BERT-BiLSTM (Bidirectional Encoder Representations from Transformers-Bidirectional Long-short term Memory) in the context of large data for online shopping. After training the unlabelled text utilising the BERT method for deep learning language, the learning textual characteristics and data provide the previously trained model for the information, which is used for extracting deeper vectors. Second, optimal textual characteristics are shown using the BiLSTM model to acquire contextual information concurrently. In the end, the BERT model and BiLSTM are combined to build a sentiment analysis model that is relevant to the customer review content. This improves the accuracy of the sentiment classification findings by merging the sentiment classification and enhancing the last vector feature.

This publication has 15 references indexed in Scilit: