상세 보기
초록
In this study, we introduced GoodGPT, a korean generative neural network system designed to enhance counseling interactions by combining retrieval-augmented generation with advanced safeguard mechanisms. By leveraging a database of real counseling records and employing Dense Passage Retrieval, GoodGPT effectively retrieves contextually relevant information to inform its generative processes. This approach not only improves the relevance and appropriateness of the generated responses but also aligns them with professional counseling fields. © 2025 IEEE.
키워드
generative model; information retrieval; mental health; neural network; safety
- 제목
- GoodGPT: Counseling-Chat
- 저자
- Kim, Kyumin; Lee, Hanyong; Lee, Jeasung
- 발행일
- 2025
- 유형
- Conference paper
- 저널명
- Digest of Technical Papers - IEEE International Conference on Consumer Electronics
- 언어
- ENG
- 출판사
- Institute of Electrical and Electronics Engineers Inc.
- ISSN
- P 0747-668X