The purpose of this study is to develop and apply an artificial intelligence chatbot for early screening and protection of students who are victims of school violence. To develop a chatbot, artificial intelligence modeling is performed by collecting cases of school violence based on natural language processing. Cases of school violence are collected online and offline to form a data dam, and LDA-based topic modeling is performed to derive keywords by type to extract intent. The chatbot engine responds according to the user's input word based on the intent classification model and the entity name recognition model. It implements an online-based chatbot system that analyzes the chatbot user's conversation content through a deep learning model and diagnoses whether or not there is a victim of school violence and the type of damage. In order to develop a chatbot, based on human-centered counseling, the flow through which counseling can be performed through three methods of empathic understanding, unconditional respect, and genuineness for the client is presented. Design a chatbot using a chatbot development API and a chatbot using a GPT model based on RNN deep learning. Comparing the two chatbots' designs, the effectiveness is analyzed through expert verification, and the research results are presented based on characteristic cases. The study result compared the contents of intent classification according to the input based on the entity and the contents of the output from flow control through sentence learning.

Authors:
Seungki Shin, Seoul National University of Education, South Korea

About the Presenter(s)
Professor Seungki Shin is a University Assistant Professor/Lecturer at Seoul National University of Education in South Korea

See this presentation on the full schedule – Schedule

Bookmark