3,206 expertly-labeled news samples in the Filipino language, consisting of 1,603 real and 1,603 fake news articles. This corpus provides a balanced dataset for binary classification tasks in a low-resource linguistic context.
Use Cases
- Train a binary classification model to detect misinformation using the news samples and their corresponding labels
- Benchmark the performance of transformer-based models on low-resource Filipino text classification
- Analyze linguistic patterns and stylistic markers that differentiate real news from fake news in Filipino
Strengths
- 3,206 total news samples with a 50/50 split between real and fake classes
- Expertly-labeled ground truth for high-quality misinformation detection
- Specifically targets the Filipino language, categorized as a low-resource domain in NLP
- Balanced distribution of 1,603 real and 1,603 fake news entries