Architecture Design for a Smart Chatbot to Enhance English Conversation Skills Using RiveScript, NLP, and ChatGPT
Dwijoko Purbohadi · 2024
This paper presents a chatbot architecture design for K-12 English-speaking learners. English learning chatbots have many advantages, including 24/7 availability, interactive interaction, instant feedback, vocabulary training, progress measurement, and effective repetition. Key technologies commonly used to build learning chatbots include natural language processing (NLP), artificial intelligence (AI), chatbot frameworks, and programming languages. The fact that chatbots can only respond to programming and data is a drawback. Chatbot engines cannot think creatively. ChatBot also needs to provide deep insights. The learning material can adapt the necessary chatbot skills to the learning context. Chatbots must match the subject matter and conversation ethics, adjust to students' abilities, and have a knowledge base. We designed a chatbot with a unique architecture and model consisting of RiveScript, NLP, ChatGPT, speech-to-text, and text-to-speech converters to meet this need. This combination results in a chatbot that can communicate with students according to the learning context.