Gardado en:
| Autor Principal: | |
|---|---|
| Formato: | Recurso digital |
| Idioma: | |
| Publicado: |
Zenodo
2026
|
| Subjects: | |
| Acceso en liña: | https://doi.org/10.5281/zenodo.18738549 |
| Tags: |
Engadir etiqueta
Sen Etiquetas, Sexa o primeiro en etiquetar este rexistro!
|
Table of Contents:
- <p><span lang="EN-US">In recent years, cybersecurity threats have become increasingly prevalent worldwide, with scam messages especially those delivered via text and online platforms emerging as a major concern. These messages aim to trick recipients into disclosing personal or financial information, leading to identity theft, privacy infringements, and financial loss. In Tanzania, the rapid growth of mobile communication technologies and mobile money services has intensified this threat. Kiswahili, being the dominant language in digital communication, adds a unique dimension to the problem. This study develops Natural Language Processing (NLP)-driven Machine Learning (ML) models to detect scam messages written in Kiswahili. A labeled dataset was constructed and models such as Support Vector Machine (SVM), Random Forest (RF), Bi-LSTM, and Swahili-BERT were evaluated. Results indicate that Swahili-BERT achieved the highest detection accuracy. The study highlights the importance of context-aware, language-specific NLP tools and recommends real-time, user-friendly, and privacy-compliant deployment strategies.</span></p>