Adaptive Machine Learning Chatbot for Code-Mix Language (English and Hindi)

Abstract

The humanoid assistant system can be described as the system resembling or imitating the human behaviour. These systems can be called as chatbots. There are a large number of conventional scripted types of chatbots. The problem with these chatbots is that they provide a monotonous type of communication i.e. they provide the user with a predefined set of options for any of its query. This scripted nature limits the scope of the chatbot systems, to provide smart and effective services to the users. This problem restricts the system efficiency. Efforts are being made to improve the scripted nature of chatbots and enable them to converse in a manner similar to the conversation between two humans. This makes the system more user-friendly, and provides better solutions to them. Chatbots providing health care services imitate the conversation between the doctor and the patients to give them general information about diseases, remedies, precautions, etc. and also provides a prediction of the diseases depending upon the symptoms provided by the user. Here, the chatbot behaves as a virtual doctor. This can be achieved by incorporating NLU, ML and NLG techniques in the system. Here, in this paper, we have briefed about the chatbot system architecture and adaptive self-learning algorithm for providing services in healthcare domain.