Built an ensemble model to detect fake reviews on Yelp with 88 accuracy and developed a question answering framework for user queries using SQuaD.

• Used models like CNN, Gradient Boosting, BERT to classify the reviews

• Used BERT trained with SQuaD dataset to build a question answering framework on genuine reviews

• Built a city heatmap of the density of the fake reviews in cities like New York and Chicago

More details about the project can be found on the github repository