Assessing the Dewey Classifier5
DESIDOC Journal of Library & Information Technology · 02.07.2026 · 46 (4) · 419-428
DOI: 10.14429/djlit.21891 ↗Kurz gefasst
Fachartikel in der Zeitschrift DESIDOC Journal of Library & Information Technology, online erschienen am 02.07.2026. Verfasst von Riya Jain und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 2.5).
Abstract (Originalsprache)
This study evaluates the effectiveness of ChatGPT Dewey Classifier5. It is an artificial intelligence tool that is used to generate Dewey Decimal Classification numbers based on bibliographic data. The main objective of this research is to determine the accuracy and consistency under different conditions. A task-based experimental method was used. It involves three validation phases, namely (A) different user IDs on the same system, (B) different user IDs on different systems at the same time, and (C) different user IDs on diferent system at different times. The 90 documents were randomly chosen among six broad subjects: Generalities, Social Sciences, Pure Sciences, Applied Sciences (Technology), Literature, and Arts & History. They were available in four languages, including Hindi, Sanskrit, Urdu and English and in various time periods, starting as early as 500 BCE up to 2020. The assessment involved four metrics, which were Exact Match, Match-Level Weighted Score, Full Consistency and Main Class Consistency. These findings indicate that the tool is reliable in broader classification, because the Main Class Consistency is 91.48 % on average, and the Weighted Match Score of 79.91 %. However, in detailed classification, the accuracy declines with Exact Match 46.30 % and Full Consistency 67.41 %. The tool performed better with Pure Science (57.78 %), but not as well with Literature (33.33 %), and Arts and History (35.56 %), presumably due to interpretive and cultural differences of those subjects. Overall, the findings indicate that, although the Dewey Classifier understands well how subjects are structured in DDC, it remains a bit ambiguous when it comes to detailed classification. It is not yet in a position to replace professional classifiers, but when employed as an aid, it can help in making classification faster and more consistent. The research suggests that a collaborative method with AI and humans is more likely to be more precise and culturally appropriate in the organisation of knowledge.
Zitieren
APA
Jain, R., & Meera (2026). Assessing the Dewey Classifier5. DESIDOC Journal of Library & Information Technology, 46(4), 419-428. https://doi.org/10.14429/djlit.21891
Vancouver
Jain R, Meera. Assessing the Dewey Classifier5. DESIDOC Journal of Library & Information Technology. 2026;46(4):419-428. doi:10.14429/djlit.21891
BibTeX
@article{Jain2026,
author = {Jain, Riya and Meera},
title = {Assessing the Dewey Classifier5},
journal = {DESIDOC Journal of Library & Information Technology},
year = {2026},
volume = {46},
number = {4},
pages = {419--428},
doi = {10.14429/djlit.21891}
}
EUVIMED