Fake Review Detector Summary

WHAT IT DOES

The Fake Review Detector Summary scans product reviews and reviewer metadata on any shopping page, then uses AI to identify patterns indicative of fake, incentivized, or otherwise unreliable reviews. It generates a comprehensive report including a trust score from 0 to 100, a detailed list of specific red flags observed (such as repetitive phrasing, sudden bursts of reviews, lack of specific details, suspiciously similar reviewer names, and inconsistent sentiment), and a one‑sentence final verdict on whether the product's reviews appear authentic. This gives you an at‑a‑glance assessment of review trustworthiness before you make a purchase decision, helping you avoid products whose ratings are artificially inflated or manipulated.

WHERE IT RUNS

It runs as a draggable overlay panel inside the BotGentz browser extension on any website that displays product reviews—Amazon, Best Buy, Walmart, Newegg, AliExpress, Etsy, eBay, Shopify stores, and countless others. It works on both product detail pages and aggregate review sections, automatically detecting review content regardless of the site's specific HTML structure.

HOW TO USE

1. Install the free BotGentz extension (see PLEASE NOTE).

2. Navigate to any product page that has customer reviews.

3. Click the BotGentz icon and select "Fake Review Detector Summary" from the tools menu.

4. The panel opens. Click the "Analyze Reviews" button.

5. Wait 3–7 seconds while the AI scans and analyzes all visible reviews.

6. Review the generated report: trust score (with color‑coded indicator), list of red flags, and final verdict.

7. Use the "Copy Report" button to copy the full analysis to your clipboard.

8. Drag the panel to reposition it, resize it, collapse it, or press Escape to close.

THE ACTUAL MECHANISM — multi‑feature anomaly detection with temporal and semantic divergence scoring

Unlike simple sentiment tools that just count positive vs. negative words, this detector employs a four‑dimensional anomaly scoring system. First, it performs semantic divergence analysis by computing cosine similarity between review texts—fake review farms often reuse templates, producing texts with abnormally high similarity scores (>0.85) across multiple reviews. Second, it applies temporal burst detection: if reviews are timestamped (or can be inferred from page order), it flags clusters of more than five reviews appearing within a 24‑hour window as a potential orchestrated campaign. Third, it runs a lexicon‑based superlative density check, counting excessive absolute terms ("amazing", "perfect", "best ever", "life‑changing") relative to the total review length; legitimate reviews typically use moderated language. Fourth, it performs reviewer metadata consistency scoring—if reviewer names are generic ("John", "Mary", "User123") or follow a detectable pattern (e.g., all start with the same letter), this increases the suspicion score. These four metrics are then weighted and combined into a final score (0‑100), with thresholds for low (0‑39, suspicious), medium (40‑69, mixed), and high (70‑100, likely authentic). The AI layer then translates these quantitative scores into plain‑English red flags and a clear verdict, making the technical analysis accessible to any shopper.

THE PANEL

The panel is fully draggable by its title bar and snaps to the nearest edge of the viewport when released. You can resize it horizontally by dragging the left or right border; the height expands to fit content up to a maximum, with scrolling inside the results area. A collapse button minimizes the panel to a small floating chip showing "⚙️ Review Detector" — click it to expand back. Press Escape at any time to close the panel entirely. The panel remembers its last position, size, and expansion state per domain via localStorage.

PLEASE NOTE

- This tool requires the free BotGentz extension (available for Chromium‑based browsers and Firefox) and an active monthly subscription to the Summarizer plan.

- AI‑generated fake review analyses are for informational guidance only—always use your own judgment and consider multiple sources (e.g., professional reviews, video unboxings, and return policies) before making a purchase decision.

- Review sections and their HTML structures vary widely and change frequently; if the detector fails to find reviews or produces an incomplete report, use the "Report issue" link in the panel footer. We monitor reports and push script updates regularly.

- No review content, product data, or personal information is ever stored on our servers. Page text is transmitted securely to the AI endpoint solely for processing and is not retained. Only anonymized usage metrics (panel open/close, analysis runs, error reports) are collected; you can disable this in the extension settings at any time.