Long Short-Term Memory based Conversation Modelling

Conversation models are becoming more popular with every day. Major companies like Apple, Microsoft, Google and Amazon have invested major funds and time and effort into such models. However, at present, these conversation models are predominantly voice chat. This does not create much difference because all the algorithms and methodologies all of the deep natural language processing techniques are all applicable to both text and voice. Our model predicts the next sentence given the previous sentence. We have developed a neural network-based model to answer questions in the form of generating sequences about the data it has been trained upon. Our model makes use of two popular neural network architecture: Sequential Learning using Recurrent Neural Networks (RNN) and Long Short-Term Memory (LSTM).