Direct answer
A plagiarism detection tool is software that compares a submitted document against large indexed collections of text and reports the passages that match, usually as a percentage plus a source-by-source breakdown. It does not decide whether you plagiarised; it surfaces evidence that a human — normally your instructor — interprets in context [1]. The most widely used tool in universities across the US, UK, Canada, Australia, New Zealand, and Ireland is Turnitin, and its report is the one most students are actually trying to anticipate. Understanding what a tool compares, why two tools disagree, and whether you can preview the real report is what separates a useful check from a false sense of security.
Introduction
A plagiarism detection tool is software that compares a submitted document against large indexed collections of text and reports the passages that match, usually as a percentage plus a source-by-source breakdown. It does not decide whether you plagiarised; it surfaces evidence that a human — normally your instructor — interprets in context [1]. The most widely used tool in universities across the US, UK, Canada, Australia, New Zealand, and Ireland is Turnitin, and its report is the one most students are actually trying to anticipate. Understanding what a tool compares, why two tools disagree, and whether you can preview the real report is what separates a useful check from a false sense of security.
How Does a Plagiarism Detection Tool Actually Work — What Does It Compare My Paper Against?
A plagiarism detection tool works by breaking your document into text segments, hashing or fingerprinting those segments, and looking for matches inside indexes the vendor has built or licensed. Turnitin's comparison runs against three distinct source families: the public internet, a subscription collection of journals and publications, and its student paper repository [2]. Your similarity score is simply the share of your document that matched something in one of those three pools.
The student paper repository is the part students understand least. It contains papers previously submitted to Turnitin by participating institutions — not every essay ever written, and not your own unpublished draft unless it has been submitted somewhere that stores it [2]. That is why a passage can look original to you and still match: an earlier cohort submitted a near-identical paragraph two years ago, and that paper is in the index.
Matching is fundamentally string- and pattern-based, which cuts both ways. Direct quotation, block quotes, and reference lists inflate the percentage even when they are correctly cited, while heavy paraphrasing that preserves the original sentence structure can still be flagged [2]. Genuine rewriting that changes both wording and syntax is what actually moves the number.
The output is a report, not a verdict. Turnitin presents a similarity index with every match highlighted and linked to its source, so a reader can open each one and judge whether it is a legitimate citation, a common phrase, or an actual problem [1]. A 28% score built from a long bibliography and three quoted definitions is a very different document from a 28% score built from an uncited paragraph.
Why Do Different Plagiarism Detection Tools Give Me Different Similarity Scores for the Same Document?
Because each tool searches a different index with different settings, the same essay routinely produces different percentages on different platforms. A checker with no journal subscription cannot match what it never licensed, and a checker with a small repository cannot match a paper submitted elsewhere [3]. The index, not the algorithm, is usually the bigger variable.
Settings move the number just as much. Most tools let the operator exclude quotations, exclude the bibliography, exclude matches below a word-count threshold, or exclude a small percentage of total matches — and each toggle changes the reported figure without changing a single word of your text [3]. Two students submitting identical files to the same system can therefore receive different scores if their instructors configured the assignment differently.
There is also no universal "safe" threshold. Turnitin does not publish a pass mark, and institutions set their own expectations, often interpreting a high percentage as legitimate when the matched material is properly attributed [3]. Treating "under 20%" as a target is a habit students invent, not a rule any detection vendor publishes.
The practical implication is that a number from an unknown checker tells you very little. What matters is whether the tool you used mirrors the index, the exclusion settings, and the report layout that your own institution will apply — otherwise you are optimising against a different exam.
Can I See the Same Similarity Report My Professor Will See Before I Submit?
Sometimes, but usually not by choice. Student access to Turnitin normally comes through a course assignment set up by an instructor, not through a personal account, so whether you can self-check depends entirely on whether your module enabled a draft or practice dropbox [4]. Many courses do not, which leaves students guessing at the one number that matters most.
When a draft submission is permitted, the report generated on the draft is produced by the same engine and the same settings as the final submission, so the two reports correspond closely [4]. The caveat is the repository setting: depending on how the assignment is configured, a submitted paper may be stored and compared against future submissions, and some institutions deliberately use a non-repository setting for drafts [4]. Knowing which setting applies to you is part of reading the report honestly.
The gap most students fall into is using a free or generic checker, getting a reassuring number, and assuming it transfers. It does not, because that tool almost certainly lacks the journal index and the student repository that produced the institutional figure [3]. A pre-submission check is only informative if it reproduces the same source pools and the same exclusions.
That is the specific problem turnitin0 was built to solve: it produces the AI detection report and the similarity report together, in the format instructors see in their learning management system, on your own draft, before you commit to a final upload.
If you have read this far, you already know the risk: a generic checker's percentage is not the percentage that decides your grade. Turnitin0 lets you see the actual report — the similarity index, every highlighted match, the source list, and the AI writing score — on the draft you are still free to edit. Documents are checked without being added to the student paper database, English files between 300 and 30,000 words are accepted, and most reports return in under 15 minutes.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is a high similarity score the same as plagiarism?
No. The similarity index measures matched text, and correctly cited quotations, reference lists, and common academic phrasing all raise it legitimately [1]. Instructors read the underlying matches, not the headline number.
Which databases does Turnitin check against?
Three pools: the public internet, a licensed collection of journals and publications, and the student paper repository containing work previously submitted by participating institutions [2]. A tool missing any one of these will under-report matches.
Can I run a plagiarism check without my paper being stored?
Yes, if the service uses a non-repository setting. Turnitin0 checks your file without adding it to the student paper database and does not share reports with third-party databases, so a draft check does not become a future match for someone else [4].
Why is my score different from a friend's on the same assignment?
Most likely different exclusion settings — quotes, bibliography, or small-match thresholds — or a different index entirely [3]. Confirm which settings your assignment uses before comparing numbers.
Does a plagiarism detector also catch AI-written text?
Modern Turnitin reports include a separate AI writing indicator alongside the similarity index, and the two are scored independently [1]. A low similarity score says nothing about whether the prose reads as machine-generated.