FAKE REVIEW DETECTION USING HYBRID BILSTM AND CNN DEEP LEARNING MODEL ON MULTI-DOMAIN TEXTUAL DATA
Keywords:
Fake Review Detection, Deep Learning, BiLSTM, CNN, Text Classification, Natural Language Processing (NLP), SHAP, Explainable AI, Multi-Domain Classification, Model InterpretabilityAbstract
Fake reviews distort consumer trust and mislead online buyers. This study presents a hybrid deep learning framework combining Bidirectional Long Short-Term Memory (BiLSTM) and Convolutional Neural Networks (CNN) to detect fake reviews using textual data from multiple domains including e-commerce, hotels, and services. The model captures both long-range contextual and local phrase-level patterns from review texts. Training and evaluation on a large multi-domain dataset demonstrate strong detection performance with accuracy of 92.7%, F1-score of 0.927, and AUC of 0.9805. SHAP explainability techniques provide model interpretability, illustrating important textual patterns influencing predictions. This approach shows promise as an effective, scalable, and interpretable solution for fake review detection based primarily on review text analysis
