Project Abstract

This project aims to lighten the burden of Swansea University’s IT Team by building a chatbot they can use for providing a first line of support. This is important because a chatbot can cater to many users at once, can provide support all day and any day, and may be easy to upgrade in the future. These potential benefits are my motivation for building a chatbot for the IT Team.A unique aspect of this project is to use Large Language Models (LLMs), or some other neural models, instead of rule-based models. The popularity of LLMs exploded in November 2022 with the public release of ChatGPT, and this author plans to evaluate different LLMs (LLaMa 3, Falcon, Mixtral etc) and use the most suitable one for different parts of the chatbot (intent recognition, response generation etc). The author also plans to use tools like LangChain, FastAPI and possibly React to build this chatbot. The project will use data from past IT queries or IT FAQs (frequently asked questions) as resources for its creation and testing.

Keywords: chatbot, Large Language Model, Machine Learning

 Conference Details

Session: Presentation Stream 22 at Presentation Slot 3

Location: CoFo 002 at Wednesday 8th 09:00 – 12:30

Markers: Deepak Sahoo, John Tucker

Course: MSc Computer Science, Masters PG

Future Plans: I’m looking for work