EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Integrating Support Vector Machine Classifiers for Real-Time Sentiment Analysis and Fraud Detection in A Fashion E-Commerce Platform

O. Nasirudeen Kazeem, Lawal Qozzim Eniola, Musa Mohammed, Usman Ahmad Tijani

Scientific Journal of Engineering, and Technology · 19.07.2026 · 3 (2) · 25-31

DOI: 10.69739/sjet.v3i2.2052 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Scientific Journal of Engineering, and Technology, online erschienen am 19.07.2026. Verfasst von O. Nasirudeen Kazeem und 3 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Fashion e-commerce platforms deal with two persistent problems that rule-based systems handle poorly: fraudulent transactions and unreliable customer reviews. This study describes the design and implementation of Angel's Corner, a Django-based fashion e-commerce application that addresses both problems by embedding trained Support Vector Machine classifiers directly into the user workflow. Two separate SVM pipelines were built. For fraud detection, an RBF-kernel SVM was trained on 284,807 credit card transactions from Kaggle. Because only 0.17% of records were fraudulent, SMOTE was applied after feature scaling to bring the training distribution to a 1:10 fraud-to-legitimate ratio. The model achieved a fraud class F1-score of 0.90 at 99.9% overall accuracy. For review sentiment, a linear-kernel SVM was paired with TF-IDF vectorization and trained on 23,486 women's clothing reviews, achieving 88% overall accuracy and a macro F1-score of 0.84 across positive, neutral, and negative classes. Both classifiers surface directly in the user interface via a real-time sentiment preview badge that updates as a customer types a review, and a dynamic risk gauge on the order confirmation page that displays the fraud assessment right after checkout. Both predictions complete in under one second. The main finding is that SVM classifiers can be deployed at the user interaction layer of a working e-commerce platform without specialist hardware, which makes this kind of intelligent analysis practical for small and mid-sized fashion retailers.

Zitieren

APA

Kazeem, O. N., Eniola, L. Q., Mohammed, M., & Tijani, U. A. (2026). Integrating Support Vector Machine Classifiers for Real-Time Sentiment Analysis and Fraud Detection in A Fashion E-Commerce Platform. Scientific Journal of Engineering, and Technology, 3(2), 25-31. https://doi.org/10.69739/sjet.v3i2.2052

Vancouver

Kazeem ON, Eniola LQ, Mohammed M, Tijani UA. Integrating Support Vector Machine Classifiers for Real-Time Sentiment Analysis and Fraud Detection in A Fashion E-Commerce Platform. Scientific Journal of Engineering, and Technology. 2026;3(2):25-31. doi:10.69739/sjet.v3i2.2052

BibTeX
@article{Kazeem2026,
  author = {Kazeem, O. Nasirudeen and Eniola, Lawal Qozzim and Mohammed, Musa and Tijani, Usman Ahmad},
  title = {Integrating Support Vector Machine Classifiers for Real-Time Sentiment Analysis and Fraud Detection in A Fashion E-Commerce Platform},
  journal = {Scientific Journal of Engineering, and Technology},
  year = {2026},
  volume = {3},
  number = {2},
  pages = {25--31},
  doi = {10.69739/sjet.v3i2.2052}
}
Kostenlos starten →