Mobile tour guide application for Kampar tourism using an interactive chatbot with generative pre-trained language model

• Main Author: Chua, Feng Zhao
• Format: Final Year Project / Dissertation / Thesis
• Published: 2024
• Subjects: T Technology (General); TD Environmental technology. Sanitary engineering
• Online Access: http://eprints.utar.edu.my/6950/1/fyp_CS_2024_CFZ.pdf; http://eprints.utar.edu.my/6950/

Tourism in small, remote areas is way more uncommon than in large cities due to their hard-to-access information, making planning for a trip difficult and time consuming as tourists will have to scour the entire Internet to find places and attractions which they may or may not prefer as second-hand opinion can also be difficult to come by due to the area’s remote nature. Tour guides are trained to resolve these types of issues but can also introduce new problems for the tourists. This project aims to develop a chatbot for a mobile application which allows users to immediately request an area or attraction in one of these small, remote areas that they would like to go to vacation for, and immediately obtain an answer through its ‘command and answer’ model. Previous work has been done on tourism chatbots, however these chatbots are usually not language adaptive due to the primitive language model they used. Not only that, these chatbots, when providing suggestions, usually miss out other factors such as user’s preferences. The chatbot this project will be developing will consider user’s request alongside user’s preferences, for example, whether they prefer attractions that prefer being physically active or just sightseeing. In addition, the chatbot is also made to adapt to additional information the user could provide, such as the time and weather in Kampar and the user’s current coordinate. With the advent and popularity of ChatGPT, this advanced language model will be used in order for it to grab context from user’s messages to accurately determine user’s request with little misinterpretation, and to grab context from user reviews for tourism areas to evaluate how well should it be suggested. Web scraping is performed to feed much more specific Kampar tourism knowledge into a language model’s database which will allow it to provide more comprehensive tourism answers. The chatbot also aims to act like a tour guide and provide information regarding an attraction or site whenever users request for it. The chatbot also provides extra functions which can improve the user’s traveling experience such as providing directions, updating user’s traveling preferences in real-time and providing user’s the ability to make reviews of the places in Kampar.