A naukri.com initiative
Dev
2w
99
Image Credit: Dev
🛡️ A Swahili SMS Scam Dataset and a Machine Learning Tool to Use It
- Scammers in Tanzania are using SMS scams, impersonating trusted individuals, to deceive people into sending money.
- A Tanzanian developer created the Bongoscam dataset with over 1,500 Swahili SMS scam examples and a machine learning model to detect them.
- The dataset contains categorized messages – scam and trust examples to train models for detection.
- The machine learning model achieved 98.7% accuracy using CountVectorizer and Multinomial Naive Bayes.
- The model is wrapped in a Flask API and accessible through a public website.
- The project structure includes GitHub repositories for the frontend, backend, API example, and instructions for setup.
- An API endpoint allows users to input SMS messages for scam prediction.
- The initiative aims to enhance digital safety by open-sourcing data, making the model public, and supporting the Swahili language.
- The BongoScam dataset serves as a foundation for localized machine learning solutions in Swahili, aiding in the fight against digital fraud.
- The project encourages developers, linguists, security researchers, and students to contribute and collaborate in combating fraudulent activities in Tanzania.
- The tool and dataset are available for testing, exploration, and contribution on various platforms.
- The goal is to build AI solutions in Swahili to protect individuals from cyber threats.
Read Full Article
5 Likes
For uninterrupted reading, download the app