Evaluation of E-commerce Logistics Service Quality using Online Customer Reviews with Bi-LSTM and Bahdanau Attention

In this research, of evaluation of E-commerce logistic service quality using the customer reviews has become important prediction to satisfy the customer. Logistic services were evaluated by traditional approaches such as interviews, surveys. Machine learning (ML) methods consume more time to extract public opinions and also failed to analyze the mixed opinions with noise in the raw data. To overcome these challenges and to predict the customer's mixed reviews and Logistics service quality evaluation effectively, a DL (Deep learning) algorithm namely Bi-LSTM (Bidirectional Long Short-Term Memory) is integrated with Bahdanau Attention Algorithm (Bi-LSTM- Bahdanau Attention). The attention algorithm is employed to increase performance and signify accurate emotion. Initially, employed Yelp review dataset, pre- processing is performed through text cleaning, tokenization and padding to reduce noise. The feature extraction is performed to the obtained processed data using word embedding (Word2Vec), Term Frequency-Inverse Document Frequency (TF-IDF), which converts the text to vector data. The classification is performed using the proposed Bi-LSTM-Bahdanau Attention model and has achieved higher accuracy of 95.67%, precision 94.89% and recall is 99.54% respectively when compared to conventional methods such as LSTM and Triple Graph Neural Network (TGNN).