Advanced Fake Review Detection via Aspect Extraction and TPA-BiLSTM-CSSA

In recent years, methods to detect bogus reviews have attracted the attention of numerous businesses and educational organizations. In order for reviews to accurately represent genuine user experiences and opinions, it is quite challenging to identify fake reviews. Supervised learning has been a popular approach to tackle the problem. The issue is that manually marking fake reviews with a degree of reliability that would be useful for training is very difficult, if not impossible. There are three phases to the suggested strategy preprocessing, feature extraction, and training the model. This proposed employs a series of preprocessing techniques to get the dataset's raw data ready for computational operations. Dimensionality reduction and the quantification of content-related and text-segment-related features comprise feature extraction. To train the model, a TPA-BiLSTM-CSSA was used. On average, the suggested method achieves a higher accuracy of 93.71% than BiLSTM and CSSA.