Human-in-the-Loop Generation of Preservation Metadata for Research Dataset Segments : Poster - iPRES 2025 - Wellington, New Zealand / Te Whanganui-a-Tara, Aotearoa
Abstract
We propose a human-in-the-loop method for generating preservation metadata for research dataset segments using a conversational GPT-based assistant. Through interactive prompts, the agent engages users in documenting the content, provenance, and context of individual files. The method outputs both a human-readable markdown summary and a METS v2-compliant metadata package with embedded JPCOAR2-style descriptions. By integrating metadata creation into the research workflow itself, this approach lowers the barrier to producing submission-ready information packages while preserving interpretive and contextual knowledge essential for long-term access.Details
- Creators
- Akihiro Kameda; Tamako Kitaoka; Makoto Goto
- Institutions
- Date
- 2025-09-01 00:00:00 +0000
- Keywords
- digital humanities
- Publication Type
- poster
- License
- http://creativecommons.org/licenses/by-sa/4.0/
- Download
- 143457 bytes