A Robust Text Dependent Methodology for Automatic Detection of Product Fake Reviews using TF-IDF
Keywords:
Long short-term memory networks (LSTM), Natural language processing (NLP), Natural Language Toolkit (NLTK), Recurrent neural networks (RNN), Term frequency-inverse document frequency (TF-IDF)Abstract
A rise in fake product reviews, which companies employ to attract more customers, has been brought on by the growth of social reviews on the internet. As they only concentrate on linguistic characteristics and miss the semantic meaning of reviews, current models for the detection of fake reviews are limited in their ability to distinguish between fake and genuine reviews. To solve this issue, we provide a brand-new ensemble model that integrates transformer architecture with modelling, TF-IDF models, and false review detection. In addition, our method employs suspicion to spot phone reviewer groups. Our results demonstrate that our suggested method beats state-of-the-art baselines in identifying fraudulent reviewer groups after thorough testing on real-world datasets.
