Online reviews have become increasingly important in the world of e-commerce, serving as a powerful tool
to establish a business's reputation and attract new customers. However, the rise of fake reviews has
become a growing concern as they can skew the reputation of a business and deceive potential customers.
As a result, detecting fake reviews has become a key area of research in recent years. This project proposes
a machine learning-based approach to detect fake reviews. The method utilizes various feature
engineering techniques to extract different behavioural characteristics of reviewers, such as the length of
reviews and the frequency of review submissions. These characteristics are then used to train different
algorithms, including K-Nearest Neighbors (KNN), Random Forest, and Support Vector Machine (SVM),
to classify reviews as either genuine or fake. The proposed technique was evaluated using a real dataset
extracted from the internet, and the results showed that SVM outperformed the other classifiers in terms of
accuracy. This suggests that SVM is a powerful algorithm for distinguishing between genuine and fake
reviews. However, the study also suggests that there is potential to improve the performance of the model by
integrating more behavioural characteristics of reviewers, such as how frequently they do reviews and how
long it takes them to complete reviews. In conclusion, this project highlights the importance of detecting
fake reviews and proposes a machine learning-based approach to achieve this. The study shows that SVM
is a powerful algorithm for this task, but there is potential for further improvement by incorporating more
reviewer behavioural characteristics. The findings of this research have practical implications for
businesses, consumers, and researchers in the field of e-commerce.