Dialog prediction in Institute Admission: A Deep Learning Way

Chatbot, an application capable to conduct online chat conversation via test, text to speech or speech to text, giving an experience of providing direct contact to a human agent. A chatbot is designed to simulate the way a human talks / behaves during a conversation and require lots of tuning and testing to pass the turing test. Chatbots are used for various purposes like customer service, routing a request, information gathering, complaint registration, e-commerce, health, educations and many more. Chatbot domains can be categorized in two types, Open and Closed. To make a closed domain chatbot precise information, questions and their answers are needed for training. While making closed domain chatbots the amount of information is less compared to open domain. Chatbots can also be further divided into two types - retrieval based chatbots, which use predefined input patterns and predefined responses, and generative based which use seq2seq neural networks and generate a answer based on large amount of data. The required data for this chatbot is collected by a survey done with the help of Google forms and is cleaned using Python and Natural Language Tool-Kit. Classification of classes is done using various machine learning techniques such as Stochastic gradient, Sequential model, Rectified Linear activation function (ReLu), Softmax. 3 Layers of Neural Networks are used to create the model. The above mentioned models and functions give accuracy of outputs more than 90%.