Melola Discovery

← Melola Discovery

Open access Journal SINTA 3

EVALUATING ROBERTA AND GPT-BASED MODELS FOR SDG MULTICLASS TEXT CLASSIFICATION ACROSS DIFFERENT DOCUMENT LENGTHS

BAREKENG: Jurnal Ilmu Matematika dan Terapan · Vol. 20 No. 3 (2026) · pp. 2645–2664

DOI
10.30598/barekengvol20iss3pp2645-2664
ISSN
2615-3017, 1978-7227
Published
2026-04-08
Publisher
PATTIMURA UNIVERSITY
Language
eng
License
creativecommons.org/licenses/by-sa/4.0

Abstract

Multiclass text classification remains a difficult task, primarily due to semantic ambiguity and differences in input length. This study evaluates RoBERTa and GPT-based models for multiclass text classification, focusing on how prompting strategies and document length affect accuracy and robustness. Experiments were conducted using the OSDG Community Dataset, which contains approximately 15,000 labeled samples. The dataset was partitioned into four subsets based on input length: short, medium, long, and all combined. Three GPT variants (zero-shot, few-shot, and fine-tuned) were compared against a RoBERTa baseline. Fine-tuning was implemented via OpenAI’s supervised API with prompt-response formatting. Performance was assessed through F1-score, precision, recall, and balanced accuracy. Fine-tuned GPT achieved the strongest results in all settings, with a macro F1-score of 0.9204 on the all-combined dataset, representing a 4.61% improvement over RoBERTa. Consistent gains were also observed across short (8.63%), medium (3.83%), and long (20.31%) texts. The largest improvement occurred on long documents, while medium-length inputs provided the most stable performance across models. These findings highlight the effectiveness of task-specific fine-tuning in enhancing GPT’s capability to classify SDG-related texts across diverse input lengths.

Collaboration network of the authors

40 authors · 118 co-authorship links · 15 across different organisations · showing the 40 most connected

Hover or tap a name to highlight its co-authors. Double-click to open the author's page.

  • IPB University
  • University of Brawijaya
  • Universitas Amikom Purwokerto
  • Universitas Muhammadiyah Jakarta
  • Universiti Putra Malaysia
  • University of Bengkulu
  • Politeknik Statistika STIS
  • Different organisations
  • Same organisation
  • Affiliation unknown
  • Thicker line = more joint articles · bigger circle = more articles
Show as table
Co-authorOrganisationJoint articles
Bagus SartonoIPB University7
I Made SumertajayaIPB University4
Asep SaefuddinIPB University4
ERFIANI ERFIANIIPB University4
Hari WijayantoIPB University3
Anik DjuraidahIPB University3
Kusman SadikIPB University3
Gerry Alfa DitoIPB University2
INDAHWATI INDAHWATIIPB University2
Anang KurniaIPB University2
Khairil Anwar NotodiputroIPB University2
Alfa Nugraha PradanaIPB University2
Ade NurhopipahUniversitas Amikom Purwokerto1
Muh. SunanIPB University1
Unique Desyrre A. ResiloyIPB University1
Desy EndrianiIPB University1
Imam CholissodinUniversity of Brawijaya1
Lintang Resita MayangsariUniversity of Brawijaya1
Andhica PratamaUniversity of Brawijaya1
Leni Anggraini SusantiIPB University1
Pradnya Sri RahayuIPB University1
Nabila Ghoni Trisno HidayatullohIPB University1
Sachnaz Desta OktarinaIPB University1
Ismah IsmahUniversitas Muhammadiyah Jakarta1
Sim Hui XinUniversiti Putra Malaysia1
Andi Harismahyanti A.IPB University1
Farit Mochamad AfendiIPB University1
Kusnaeni KusnaeniIPB University1
Anggita Rizky FadilahIPB University1
Maulida Fajrining TyasIPB University1
Husnun NashirIPB University1
Ni Gusti Ayu Putu Puteri SuantariIPB University1
Kristuisno Martsuyanto KapilukaIPB University1
Indra Rivaldi SiregarIPB University1
Herlin FransiskaUniversity of Bengkulu1
Ita WulandariPoliteknik Statistika STIS1

Affiliations come from OpenAlex; journal OAI-PMH metadata carries none. Authors without an OpenAlex ID are matched by name, which can merge or split people.

Metadata source: OAI-PMH (metadata jurnal) https://ojs3.unpatti.ac.id/index.php/barekeng/oai, first harvested 5 Nov 2026, last seen 15 Nov 2026.
Original metadata (oai_dc)
<oai_dc:dc xmlns="http://www.openarchives.org/OAI/2.0/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:oai_dc="http://www.openarchives.org/OAI/2.0/oai_dc/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/  http://www.openarchives.org/OAI/2.0/oai_dc.xsd">
	<dc:title xml:lang="en-US">EVALUATING ROBERTA AND GPT-BASED MODELS FOR SDG MULTICLASS TEXT CLASSIFICATION ACROSS DIFFERENT DOCUMENT LENGTHS</dc:title>
	<dc:creator>Hasanah, Uswatun</dc:creator>
	<dc:creator>Mohamad Soleh, Agus</dc:creator>
	<dc:creator>Suhaeni, Cici</dc:creator>
	<dc:creator>Fitrianto, Anwar</dc:creator>
	<dc:description xml:lang="en-US">Multiclass text classification remains a difficult task, primarily due to semantic ambiguity and differences in input length. This study evaluates RoBERTa and GPT-based models for multiclass text classification, focusing on how prompting strategies and document length affect accuracy and robustness. Experiments were conducted using the OSDG Community Dataset, which contains approximately 15,000 labeled samples. The dataset was partitioned into four subsets based on input length: short, medium, long, and all combined. Three GPT variants (zero-shot, few-shot, and fine-tuned) were compared against a RoBERTa baseline. Fine-tuning was implemented via OpenAI’s supervised API with prompt-response formatting. Performance was assessed through F1-score, precision, recall, and balanced accuracy. Fine-tuned GPT achieved the strongest results in all settings, with a macro F1-score of 0.9204 on the all-combined dataset, representing a 4.61% improvement over RoBERTa. Consistent gains were also observed across short (8.63%), medium (3.83%), and long (20.31%) texts. The largest improvement occurred on long documents, while medium-length inputs provided the most stable performance across models. These findings highlight the effectiveness of task-specific fine-tuning in enhancing GPT’s capability to classify SDG-related texts across diverse input lengths.</dc:description>
	<dc:publisher xml:lang="en-US">PATTIMURA UNIVERSITY</dc:publisher>
	<dc:date>2026-04-08</dc:date>
	<dc:type>info:eu-repo/semantics/article</dc:type>
	<dc:type>info:eu-repo/semantics/publishedVersion</dc:type>
	<dc:format>application/pdf</dc:format>
	<dc:identifier>https://ojs3.unpatti.ac.id/index.php/barekeng/article/view/22463</dc:identifier>
	<dc:identifier>10.30598/barekengvol20iss3pp2645-2664</dc:identifier>
	<dc:source xml:lang="en-US">BAREKENG: Jurnal Ilmu Matematika dan Terapan; Vol 20 No 3 (2026): BAREKENG: Journal of Mathematics and Its Application; 2645-2664</dc:source>
	<dc:source>2615-3017</dc:source>
	<dc:source>1978-7227</dc:source>
	<dc:source>10.30598/barekengvol20iss3year2026</dc:source>
	<dc:language>eng</dc:language>
	<dc:relation>https://ojs3.unpatti.ac.id/index.php/barekeng/article/view/22463/12685</dc:relation>
	<dc:rights xml:lang="en-US">Copyright (c) 2026 Uswatun Hasanah, Agus Mohamad Soleh, Cici Suhaeni, Anwar Fitrianto</dc:rights>
	<dc:rights xml:lang="en-US">http://creativecommons.org/licenses/by-sa/4.0</dc:rights>
</oai_dc:dc>