Paste a review
Analysis runs as you type or paste. The score covers text only; the real systems weigh behaviour and arrival patterns far more heavily.
Comparison corpus for duplication (optional)
Why it reads as fabricated
Ordered by contribution to the score. Every quote is taken from your text.
The evidence, in place
The reconstruction
The same content, run backwards through the fabrication signals. Editable gaps are facts the tool refuses to invent for you.
The reconstruction appears here after analysis.
How the score is built
One formula, three stages. Everything is a deterministic function of the text you paste — no model, no sampling, no network.
1 · Decompose
Each sentence is measured on six axes: structure (length, rhythm, repetition, punctuation), lexicon (promotion, superlatives, evaluative density), discourse (specificity, temporal orientation), stance (rating–sentiment fit, friction), authorship (machine-writing cues) and duplication against a corpus you supply.
2 · Normalise by coverage
A ten-word review cannot support a lexical-diversity measurement. Rather than scoring that as safe, the signal is excluded and the score is computed over what remains, with coverage reported beside it. Short texts therefore lose precision, not truth.
3 · Invert
The rewrite applies the inverse of each signal that fired: persuasion is replaced with a reuse statement, superlatives deflate to plain adjectives, essay connectives are stripped, and the recount frame is rebuilt around explicit gaps for dates, objects, friction and resolution.
The reverse transform, signal by signal
Each rule exists because the corresponding signal fired. This is the operational answer to “what would this review look like if it were real?”
| Signal | What it caught | Reverse transform | Example |
|---|---|---|---|
| LEX-01 | Boilerplate persuasion | Replace only the recommendation phrase with the reviewer's own reuse decision; the rest of the clause survives. | |
| LEX-02 | Superlative stacking | Deflate to a plain adjective. A real thing described plainly is more credible than the same thing praised maximally. | |
| LEX-03 | Intensifier inflation | Delete intensifiers outright; they carry no information the adjective lacks. | |
| ORT-01 | Thin text | Keep only sentences with content, then restore the recount frame that carries length honestly. | |
| ORT-04 | Uniform rhythm | Split the longest sentence so lengths vary; templated prose marches in step. | one 28-word sentence → 18 + 9 words |
| ORT-08 | Emphatic punctuation | Normalise exclamation marks and all-caps shouting into statements; real acronyms are kept. | |
| DIS-01 | Evaluative, not recounting | Recast into the past tense of an event that happened, not a verdict that exists. | |
| DIS-02 | No specifics | Insert explicit gaps for date, object, friction and resolution — never invented values. | |
| STA-01 | Frictionless praise | Require the small imperfection that nearly every genuine account contains. | + The one thing that went wrong was […] |
| STA-02 | Rating ≠ sentiment | Reconcile the star rating with the valence of the words, in whichever direction is wrong. | 5 stars, complaining text → friction added |
| AIG-04 | Machine-writing cues | Strip essay connectives, add contractions, remove the “throughout the process” register. | |
| DUP-01 | Corpus duplication | Flag near-duplicate phrasing; the transform cannot fix it, because the fix is a different author. | max Jaccard 0.59 → reported, not rewritten |
Every signal the page computes
IDs are stable across the workspace and match spec/signals.csv. Weights are illustrative reconstructions, not Trustpilot's.
| ID | Signal | Measurement | Weight |
|---|---|---|---|
| ORT-01 | Token length | Banded token count; short text is a prior, not a finding. | 0.09 |
| ORT-04 | Sentence rhythm | Coefficient of variation of sentence lengths. Uniform prose is a template signature. | 0.06 |
| ORT-05 | Lexical diversity | Type–token ratio, suppressed below 20 tokens where it is mechanically high. | 0.05 |
| ORT-07 | Trigram repetition | Share of repeated 3-token shingles inside the review. | 0.05 |
| ORT-08 | Emphatic punctuation | Repeated exclamation marks and all-caps words. | 0.04 |
| LEX-01 | Promotional language | Phrase matches from a lexicon of calls to action and sales idiom. | 0.10 |
| LEX-02 | Superlatives | Density of maximal adjectives per 100 tokens. | 0.08 |
| LEX-04 | Evaluative vs concrete | Ratio of judgements to facts — numbers, names, times, domain terms. | 0.05 |
| DIS-02 | Event specificity | Numbers, currency, clock times, dates, named entities, domain terms; saturates at five. | 0.07 |
| DIS-01 | Temporal orientation | Past-tense recount markers against present/future evaluation markers. | 0.05 |
| STA-02 | Rating ↔ sentiment | Expected valence from stars against observed valence from a negation-aware lexicon. | 0.08 |
| STA-01 | Frictionless praise | High rating, long text, zero negative tokens and no explicit reassurance. Genuine accounts almost always carry one cost. | 0.05 |
| AIG-04 | Machine-text cues | Contractions, essay connectives, rhythm uniformity, first-person absence. Deliberately weak. | 0.04 |
| DUP-01 | Duplication | Max Jaccard over 3-token shingles against a supplied corpus. | 0.10 |
What this cannot tell you
A high score is not an accusation, and a low score is not a clearance. Both readings misunderstand what text can carry.
Text is the weakest layer. Published research on the same n-gram method shows 89.6% accuracy on crowdworker fakes, 67.8% on real filtered reviews, and 52–54% — chance — when a model trained on one is transferred to the other. Adding behavioural features recovers roughly 14 F1 points.
Base rates decide the false-positive count. At a 1-in-1,000 fabrication rate, a detector with 95% sensitivity and 90% specificity is wrong about 99 times for every 100 reviews it flags. No text-only score escapes that arithmetic.
The rewrite is stylistic, not forensic. It removes the markers of manufacture and restores the shape of a recount. It cannot make an invented experience true — that is why every fact it needs is left as an editable gap rather than filled in.
Non-native and terse writers pay for it. Short reviews, translated reviews and reviews from writers with limited English trip length, diversity and specificity signals at higher rates. Any deployment that treats these as fabrication evidence inherits that bias.