Automatic Classification of Subjects and Sustainable Development Goals (SDGs) in Documents with Generative AI: An Experience from the Unicamp Library System.

Saved in:

Bibliographic Details
Title:	Automatic Classification of Subjects and Sustainable Development Goals (SDGs) in Documents with Generative AI: An Experience from the Unicamp Library System.
Authors:	deu Gonçalves de Oliveira Foz, Francisco Ta¹ ffoz@unicamp.br, Martins, Márcio Souza² marciosm@unicamp.br, de Souza Neves, Alessandra Karyne Clemente³ akaryne@unicamp.br, de Carvalho Mansur, Erica Cristina⁴ ericacc@unicamp.br, Eliel, Oscar⁵ oeliel@unicamp.br
Source:	Information Technology & Libraries. Mar2026, Vol. 45 Issue 1, p1-24. 24p.
Subject Terms:	Generative artificial intelligence, Qualitative research, Academic libraries, Librarians, *Automation, Sustainability, Quantitative research, Thematic analysis, Sustainable development, Abstracting & indexing services
Abstract:	This study evaluates the effectiveness of the Artificial Intelligence for Theme Generation tool (original Portuguese acronym name: IAGeraTemas), developed with generative artificial intelligence (AI; Google Gemini), for automating thematic classification and the assignment of Sustainable Development Goals (SDGs) in documents. The methodology combined quantitative analyses (metrics of precision, recall, and accuracy) on 50 articles published by authors from the State University of Campinas (Unicamp), using classification from the SciVal database and qualitative analyses (analysis of the relevance of terms indexed by librarians from the Unicamp Library System in 40 articles available in the Unicamp Institutional Repository), comparing them with manual indexing performed by librarians. The quantitative results in SDG classification showed a recall of 0.785, while the "precision" and "accuracy" metrics were moderate. The qualitative analysis deepened the evaluation of term coherence and relevance suggested by the AI versus human indexing. It revealed the tool's potential for suggesting relevant terms and expanding concepts, but it also exposed limitations in addressing complex topics. The research, conducted as an experiment at Unicamp Library System, concludes that IAGeraTemas is a valuable auxiliary tool, complementing but not replacing manual indexing, reinforcing the importance of human expertise in validating and refining results, and emphasizing the synergistic potential between AI and information professionals. [ABSTRACT FROM AUTHOR]
	Copyright of Information Technology & Libraries is the property of American Library Association and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database:	Education Research Complete

Description
Abstract:	This study evaluates the effectiveness of the Artificial Intelligence for Theme Generation tool (original Portuguese acronym name: IAGeraTemas), developed with generative artificial intelligence (AI; Google Gemini), for automating thematic classification and the assignment of Sustainable Development Goals (SDGs) in documents. The methodology combined quantitative analyses (metrics of precision, recall, and accuracy) on 50 articles published by authors from the State University of Campinas (Unicamp), using classification from the SciVal database and qualitative analyses (analysis of the relevance of terms indexed by librarians from the Unicamp Library System in 40 articles available in the Unicamp Institutional Repository), comparing them with manual indexing performed by librarians. The quantitative results in SDG classification showed a recall of 0.785, while the "precision" and "accuracy" metrics were moderate. The qualitative analysis deepened the evaluation of term coherence and relevance suggested by the AI versus human indexing. It revealed the tool's potential for suggesting relevant terms and expanding concepts, but it also exposed limitations in addressing complex topics. The research, conducted as an experiment at Unicamp Library System, concludes that IAGeraTemas is a valuable auxiliary tool, complementing but not replacing manual indexing, reinforcing the importance of human expertise in validating and refining results, and emphasizing the synergistic potential between AI and information professionals. [ABSTRACT FROM AUTHOR]
ISSN:	07309295
DOI:	10.5860/ital.v45i1.17510