Deceptive Hotel Review Detection: An Ensemble Learning Framework

Electronic Word of Mouth (eWoM) becomes a key player for e-commerce platforms to upgrade or downgrade product or service value. Currently, reviews are one of the favorable sources for individuals to decide their opinions about products, hotels, places, transport services, food quality, and many others. The digital era yields positives and negatives; positive includes fast access to multiple people's opinions in terms of legitimate reviews, whereas negative includes the existence of deceptive reviews inside the total review. Deceptive reviews may be used to create a biased impact on humans about the brand or services. Therefore, detecting such biased reviews from online platforms is an urgent requirement. To address the issues, a weighted ensemble model was developed in this research, which was trained and tested on the gold standard dataset. The model's outcome achieved approx 5% higher accuracy for detecting deceptive reviews than baseline models.