Counterfeit Listing Detection Network

Brand-protection teams don't check every online listing by hand — a crawler scores each one against a handful of measurable signals and only escalates the riskiest for human review. This simulator recreates that pipeline in 3D: a synthetic marketplace of listings is plotted by image/logo similarity, price deviation from MSRP, and seller-trust score, then a continuously running crawler scans them at a chosen rate, scores each with a logistic risk model, and routes anything past your threshold into a quarantine zone. Live precision and recall readouts — graded against each listing's hidden true label — show the classic operating-point trade-off: a stricter threshold catches fewer genuine counterfeits accidentally, but also lets more real ones slip through unflagged.