To search, Click below search items. |
| All Published Papers Search Service |
|
|
|
Title |
Collaborative Cataloging with AI: An Experimental Study on OpenAI's ChatGPT-3.5 |
|
Author |
Safinaz Mahmoud Elroukh |
| Citation |
Vol. 25 No. 6 pp. 159-167 |
|
Abstract |
The paper introduces a proposed approach to machine cataloging through cooperation between the ChatGPT-3.5 AI model and human catalogers. The experimental method was employed. There were three samples: ChatGPT-3.5, ten catalogers, and a mixed sample. Participants were tasked with creating MARC records for twelve books following specific prompts. Speed, accuracy, and efficiency were tested. Generated records underwent both automated and human validation. Findings revealed that ChatGPT-human collaboration is more efficient than the sole ChatGPT and human groups in cataloging. Some errors were detected: invalid field formats, incorrect indicator data, and length discrepancies. The study examined the efficacy of generating bibliographic MARC records for books while excluding other types of information resources or metadata standards. While participants had positive experiences with AI collaboration, concerns were raised about potential shifts in future cataloging roles. The proposed approach helps reduce the cost of building library catalogs and frees librarians' time for other tasks. The paper reports errors in the generated records, assisting developers in refining AI algorithms. It serves as a roadmap for catalogers, indicating where AI reliance is suitable and where human intervention remains vital. The study introduces a pioneering approach and provides valuable insights for the future of cataloging practices. |
|
Keywords |
Collaborative Cataloging, ChatGPT-3.5, Artificial Intelligence, MARC21, Chatbots. |
|
URL |
