Today's review rating are static in nature and does not change based on user preferences. There are many proposed solutions for adaptive review based on user preferences. However, it is very difficult to device correct preference set for every product. In addition, preferences and not always objective and maybe interdependent on other factors. What is proposed here is an adaptive review rating schema based on user rating of existing reviews and preferences derived out of review of existing purchased products.

My proposal include a deep Neural network for modeling the prediction and reinforcement module to mimic human response. The proposed neural network model comprises of four layers: Input, two activation layers and an output layer. Two activation layers are chosen to improve the accuracy of the output prediction.