Identify deceptive hotel reviews written in English with recurrent neural network
This extension spot deceptive hotel reviews with the help of deep learning. A pre-trained LSTM model with 99% test accuracy is embedded to identify the nature of your target review.
After enable the extension, simply select the review text you would like to identify with your mouse. A small message window will show after a few seconds to display the predict result. Also there is an on/off switch in the extension popup.
The predict result including the following information:
①Sentiment: A sentimental identification of the review by the extension that can be taken as a reference of the reliability of this prediction.
②Polarity: Identification result of whether it is a genuine or deceptive review.
③Confidence: How confident does the model feel about the predicting result.
* Subjected to the pre-trained model, this extension only works well on hotel reviews.