Weight Relative Self-Attention – Bidirectional Encoding Representation Transformer for Identification of False Comments in E-Commerce

False comments are known as fake opinions are utilized for misleading people and has profited much significance currently. This is because of rapid growth in online marketing transformations like purchasing and selling. E-commerce gives facility to customers for post comments and around service or product while purchasing. In this research, Weight Relative Self-Attention – Bidirectional Encoding Representation Transformer (WRSA-BERT) for identifying fake comments in e-commerce platforms. Dataset evaluated for research is Ott, Yelp dataset and various pre-processing techniques are used for dataset. The Term Frequency – Inverse Document Frequency (TF-IDF) is utilized to extract features and identification of fake comments is performed by WRSA-BERT. Performance of proposed algorithm is analyzed through performance measure like accuracy, precision, recall and f1-score. Proposed algorithm reached accuracy 97.4%, precision 94.95, recall 93.95 and f1-score 93.3 for 2 datasets. Proposed method performed superior than other existing methods like Bidirectional Long-Short Term Memory (BiLSTM) and Convolutional Neural Network (CNN) and CNN – Long-Short Term Memory (LSTM).