Related Experiment Video
Updated: Jun 24, 2026
11:18
Closed-loop Neuro-robotic Experiments to Test Computational Properties of Neuronal Networks
Published on: March 2, 2015
10.4K
Computational Intelligence Based Recurrent Neural Network for Identification Deceptive Review in the E-Commerce
Saleh Nagi Alsubari1, Theyazn H H Aldhyani2, Sachin N Deshmukh1
1Department of Computer Science & Information Technology, Dr. Babasaheb Ambedkar Marathwada University, Aurangabad, India.
Computational Intelligence and Neuroscience
|November 28, 2022
Summary
This study developed a deceptive review detection system using a recurrent neural network, bidirectional long short-term memory (RNN-BLSTM) model. The system achieved 89.6% accuracy in identifying fake reviews on Amazon and Yelp datasets, outperforming existing methods.
Area of Science:
- Natural Language Processing
- Machine Learning
- E-commerce Analytics
Background:
- Online reviews significantly influence consumer purchasing decisions.
- Deceptive reviews mislead consumers and harm e-commerce platforms.
- Fraudsters create fake reviews for personal gain or to damage competitors.
Purpose of the Study:
- To develop and evaluate a system for detecting deceptive product reviews.
- To analyze deceptive reviews in the Amazon and Yelp e-commerce domains.
- To compare the performance of a recurrent neural network, bidirectional long short-term memory (RNN-BLSTM) model against existing methods.
Main Methods:
- Utilized Linguistic Inquiry and Word Count (LIWC) tool for feature extraction from review text.
- Engineered a deceptiveness score based on features like authenticity, sentiment, and analytical thinking.
- Applied a recurrent neural network, bidirectional long short-term memory (RNN-BLSTM) model trained on word embeddings for classification.
Main Results:
- The RNN-BLSTM model achieved a testing accuracy of 89.6% on both Yelp and Amazon datasets.
- The integration of LIWC features with word embeddings demonstrated superior performance.
- The proposed method effectively identified and classified deceptive online reviews.
Conclusions:
- The developed deceptive review detection system is effective and accurate.
- LIWC features combined with word embeddings offer a robust approach for detecting fake reviews.
- This research contributes to enhancing consumer trust and e-commerce integrity.
