Photo from SmashingMagazine.com

Fake Review Generation

Photo from SmashingMagazine.com

Fake Review Generation

  • To prove that public opinion of businesses can be easily manipulated by fake reviews, we generated a topic model which generates restaurant reviews based on real restaurant reviews.
  • We aggregated, cleaned and developed a corpus of restaurant reviews from Yelp.
  • Designed Marcov, LSTM and RNN deep learning models to train on the texts and predict the upcoming texts or characters.
  • Obtained the most realistic reviews using the LSTM model.
Deep Learning Un-Supervised Learning TensorFlow Keras Marcov Model Text Analytics GloVe Embedding LSTM Convolutional Neural Networks Recurrent Neural Networks Python Pandas Matplotlib
Avatar
Sandeep Gunda
Aspiring Data Scientist