Sinhala Chatbot for Train Information

Abstract

A chatbot is an essential point at the natural evolution of a Question Answering system in parallel to Natural Language Processing (NLP) improvements. Chatbot applications enhance the interactions between people and services with enhancing customer experience. More accurate chatbot systems offer companies new opportunities to improve the customer‘s engagement with the service by reducing the typical cost of customer service. This research is on the design and implementation of the domainspecific Sinhala Chatbot System that can communicate between computer and user, through the Sinhala language. The entire system has been developed using the JAVA programming language. The main objective behind this approach was to understand user-required information and provide the ideal answers considering various user circumstances. The proposed framework includes an NLP service and a context manager. NLP service contains Intent classifier, Entity identifier and Response generator, implemented for Sinhala Language processing with Supervised Learning in Neural Network, trained with the data, collected from the end users. We have introduced a framework that can use for the implementation of a chatbot in Sinhala for any domain. By using the proposed framework, we present a simple chatbot system which accepts user input in Sinhala and generate replies extracted from the training data set in the domain of Train information in the Sinhala language.