Build of Mobile Chatbot System Based on Generative Pre-trained Transformer 3 with OpenAI

다국어 초록 (Multilingual Abstract)

Currently, there are many Chatbot applications deployed on mobile devices as well as on website platforms, chatbot technologies can be integrated into the system such as Facebook chatbot, and RASA chatbot, and they contribute many benefits in supporting Helping businesses take care of customers as well as handle problems. In addition to the strong growth of machine learning, many technology companies have built machine learning platforms that developers can use through released APIs. Specifically, OpenAI developed a GPT-3 language model with training data of about 175 billion words. At the same time, the company has published the API through a beta version, providing a lot of services using this language model, including Chatbot. The article researches using the APIs of the GPT-3 model and Google AI to build a mobile chatbot system using text and voice processing. We provide class models, and algorithms written in Java in Android operating system to invoke APIs, the other researchers can continue to extend and develop these class models and algorithms.

목차 (Table of Contents)

• Abstract
• Ⅰ. Introduction
• Ⅱ. Related Research
• 1. Research on Chatbot
• 2. OpenAI and Generative Pre-trained Transformer
• 3. Google AI Speech-to-Text
• Ⅲ. Proposal And Implementation of Research Model
• 1. Research Motivation
• 2. Computer and Mobile Environment
• 3. Proposal for Research Model
• Ⅳ. Experimental and Results
• Ⅴ. Conclusion
• References