Development of a Parenting Consultation Chatbot Utilizing LLM-Based Question Intent Estimation

Abstract

In the dynamic field of service innovation using Large Language Models (LLMs), chatbots for parenting advice are pivotal yet underexplored. Despite the demand for child-care support, few solutions effectively leverage LLMs like GPT-3.5. Our project introduces a chatbot designed to navigate the complexities of parenting queries. Initial assessments highlighted issues in precisely responding to user inquiries, often misinterpreting user intent or offering verbose answers. So we proposed a system focusing on the user's question intent. The proposed system was able to generate focused and concise responses by better identifying the user's intentions, demonstrating the potential of LLM-based chatbots in the child-care consultation field. However, issues such as the handling of complaints still remain, and further improvement of the child-care support chatbot is needed.