Detect Fake Reviews
in English & Arabic
Paste any product review below. The system automatically detects the language and classifies it as genuine or fake using fine-tuned transformer models.
98.1%
English F1
85.8%
Arabic F1
2
Languages
0
Analysed
0 / 5000
Try an example:
Confidence
English Model
distilbert-base-uncased fine-tuned on 40,432 Amazon reviews.
Achieves F1 = 0.9813 on the held-out test set.
Input: raw text up to 256 tokens.
Arabic Model
CAMeL-Lab/camelbert-mix fine-tuned on 1,728 multi-domain Arabic reviews.
Achieves F1 = 0.8582. Handles MSA and dialectal Arabic.