01
STEP 1Enter Review
Paste a product review, add a product name, and optionally include its star rating.
Input: review text + optional product context
UNDER THE HOOD
A clear process. Not a black box.
Follow a review from raw text to an explainable prediction, one step at a time.
Demo mode
02
STEP 2Text Processing
The system preprocesses the text and examines its structure, words, punctuation, and length.
Normalize text → tokenize → inspect patterns
03
STEP 3ML Model Analysis
A connected machine-learning model would examine linguistic and behavioral patterns to estimate authenticity.
Currently: transparent text-based demo heuristics
04
STEP 4Prediction
View a predicted classification, an illustrative confidence score, and supporting indicators.
Output: classification + probabilities + explanation
THE ANALYSIS PIPELINE
From words to insight.
A modular architecture, ready for a trained ML backend.
ReviewStage 01
PreprocessingStage 02
Feature ExtractionStage 03
ML ModelDemo heuristic today
PredictionStage 05
What the demo looks for
The local heuristic combines several signals. None is conclusive on its own.
- Balanced opinions and concrete product details
- Promotional wording and repeated expressions
- Capitalization, punctuation, and sentiment intensity
- Review length and rating / text consistency
What it cannot tell you
A convincing review can be fabricated. A strongly worded review can be sincere.
- The demo is not a trained or validated ML model.
- Scores are illustrative, not calibrated probabilities.
- Reviewer history and purchase verification are unavailable.
- Results are signals for reflection, not proof of fraud.
Ready to see the process in action?
Try a sample, or bring a review of your own.