Implementation of Insurance Chatbot on Multiple Platform

Tan, Chee Onn (2023) Implementation of Insurance Chatbot on Multiple Platform. Final Year Project (Bachelor), Tunku Abdul Rahman University of Management and Technology.

Text RDS_Tan Chee Onn_Fulltext.pdf Restricted to Registered users only Download (3MB)

Abstract

Purpose: The main purpose of this project is to build an insurance chatbot which is engaging to attract insurance customers to use it and provide convenience and save the cost for the insurance company. The existing insurance chatbot is not engaging since it cannot carry out a conversation with the user. Scope: The main function of the insurance chatbot is to collect the user's personal information and requirements to let the insurance agent perform the tasks easily. It is also able to answer simple questions asked by users. If it faces a complex question, it will pass to a human customer service agent to solve the questions. The insurance chatbot is used in the insurance industry. Methodology: The methodology used is the waterfall model and IBM Watson Assistant framework. The waterfall model includes 6 sequential phases which are requirement analysis, system design, implementation, testing, deployment and maintenance. The requirement-gathering technique used is an online questionnaire. The convenience sampling method is used in the questionnaire. Testing criteria used: The testing stages used are Unit Testing, System Testing and User Acceptance Testing. Unit Testing uses the white-box testing technique, whereas Unit Testing and System Testing use the bottom-up and black-box testing techniques. Results and Conclusion: The insurance chatbot is successfully implemented on different platforms. The insurance chatbot is able to answer simple queries asked by customers in an interactive way and lessen the burden of human customer representatives by helping them to collect customers’ personal information to perform some tasks required by customers. The insurance chatbot does not connect to the database. Therefore, it cannot retrieve the data from the database directly to answer customers’ queries. Besides, the insurance chatbot does not apply Watson Speech to Text service. Hence, customers cannot enter commands by talking.