EUVIMEDEUVIMED entdecken
Crossref

Detecting complaints on e-commerce websites using large language models

Phung Kim Thai, Nam Son Tran, Thuy Ngoc Thi Do

Journal of Development and Integration · 15.07.2026 · (88) · 56-63

DOI: 10.61602/jdi.2026.88.07 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Development and Integration, online erschienen am 30.06.2026. Verfasst von Phung Kim Thai und 2 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In the context of the rapid proliferation of e-commerce, customer comment on online platforms, particularly complaints, holds significant value for businesses. Although traditional machine learning approaches are prevalent in text classification, they present certain limitations that hinder their application in tasks such as complaint detection. Therefore, this study proposes a novel approach that utilizes Large Language Model (LLM) for the task of e-commerce complaint detection without requiring complex fine-tuning. The results indicate that LLM achieve high performance, with average Accuracy and F1-scores exceeding 0.90 in most cases. This study contributes to expanding the application scope of LLM in Vietnamese language processing and provides a valuable tool for enterprises to automatically identify issues from customer comments.

Zitieren

APA

Thai, P. K., Tran, N. S., & Do, T. N. T. (2026). Detecting complaints on e-commerce websites using large language models. Journal of Development and Integration, 56-63. https://doi.org/10.61602/jdi.2026.88.07

Vancouver

Thai PK, Tran NS, Do TNT. Detecting complaints on e-commerce websites using large language models. Journal of Development and Integration. 2026:56-63. doi:10.61602/jdi.2026.88.07

BibTeX
@article{Thai2026,
  author = {Thai, Phung Kim and Tran, Nam Son and Do, Thuy Ngoc Thi},
  title = {Detecting complaints on e-commerce websites using large language models},
  journal = {Journal of Development and Integration},
  year = {2026},
  number = {88},
  pages = {56--63},
  doi = {10.61602/jdi.2026.88.07}
}
Kostenlos starten →