Szczegóły publikacji
Opis bibliograficzny
Assessing AI techniques on sentiment detection in scholarly peer reviews: insights from BERT-masking, ChatGPT, and Python classifiers / Pascal Muam MAH, Iwona SKALNA, Tomasz PEŁECH-PILICHOWSKI // W: EdMedia + Innovate Learning [Dokument elektroniczny] : World Conference on Educational Media & Technology : May 19–23, 2025, Barcelona, Spain. — Wersja do Windows. — Dane tekstowe. — Waynesville : Association for the Advancement of Computing in Education (AACE), [2025]. — e-ISBN: 978-1-939797-83-4. — S. 1415–1420. — Wymagania systemowe: Adobe Reader. — Tryb dostępu: https://www.learntechlib.org/p/226365/ [2025-07-29]. — Bibliogr. s. 1419–1420, Abstr. — Dostęp po zalogowaniu
Autorzy (3)
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 161460 |
|---|---|
| Data dodania do BaDAP | 2025-08-08 |
| Rok publikacji | 2025 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp |  |
| Konferencja | World Conference on Educational Multimedia, Hypermedia and Telecommunications 2025 |
Abstract
This study explores the role of natural language processing (NLP) and emotional intelligence in analyzing scholarly peer review sentiments using two approaches: ChatGPT and a Python-based emotional tone classifier. With the growing influence of artificial intelligence, one persistent challenge in academia is addressing emotional fragility and dissatisfaction among reviewers and editorial boards. To investigate this, the researchers analyzed a dataset of peer review interactions from Harvard Dataverse using Google Colab for Python-based analysis and ChatGPT 3.5. The Python method offered a reliable, standardized approach for detecting emotional sentiment, making it a strong baseline. In contrast, ChatGPT demonstrated a broader, more generalized understanding of emotions within academic discourse, although it lacked percentage-based sentiment scores. Results showed Python’s extensive performance in data-driven tasks, while ChatGPT offered nuanced emotional interpretations based on its diverse training data. In conclusion, Python provides a dependable method for emotional analysis in scholarly reviews, whereas ChatGPT delivers versatile and context-rich insights. Both approaches contribute significantly to improving scholarly review processes, each with unique strengths in emotional sentiment analysis.
