Direct answer
A similarity detection system is software that compares a submitted document against a large indexed collection of other texts — websites, academic publications, and previously submitted student papers — and reports the percentage of the document that overlaps with those sources. In Turnitin, that output is called a Similarity Report: a percentage plus a list of numbered, colour-coded sources with the matching text highlighted [1]. The percentage measures overlap, not misconduct; it is evidence to be read, not a verdict to be feared [1].
How Does a Similarity Detection System Actually Work?
A similarity detection system works by fingerprinting text rather than by "understanding" it. When a document enters the system, the text is broken into sequences of words that are matched against an index built from the internet, subscription publication databases, and — depending on the institution's settings — a repository of student papers previously submitted to that system [2]. Because the comparison is mechanical, the system will flag a correctly quoted paragraph and a copied paragraph identically; the difference is only visible when a human reads the surrounding citation. This is why the report is designed to be interpreted, not simply scored.
The repository question matters more than most students realise. Whether your paper is added to the standard student paper database is controlled by the institution's account settings, not by the detector's design, and many institutions run non-repository checks so that drafts are compared against the corpus without being retained in it [2]. That distinction is the difference between a rehearsal and a performance: a non-repository check gives you the same comparison against the same corpus, but your draft does not become a future match for someone else.
Processing is fast because the index is pre-built. Turnitin-style systems return a report within minutes of submission, and the report is delivered to the instructor through the learning management system rather than emailed as a standalone file [2]. For students, the practical consequence is that a report can appear on an assignment almost immediately after the deadline — there is no quiet window in which a flagged draft can be quietly replaced.
What Does a Similarity Score Percentage Actually Mean?
The similarity score is the proportion of a submission's text that matches sources in the system's index — nothing more, and nothing less [3]. It is not a plagiarism percentage, and it is not additive in the way students assume: if the same sentence appears on three websites, it counts once toward the percentage but is listed as three separate sources, so the source percentages in a report can sum to more than the headline score [3]. Reading the number without reading the matches is the single most common misinterpretation of a similarity report.
Report settings change the number you see. Instructors can exclude quoted material, exclude the bibliography, and set a small-match threshold that hides matches below a chosen word count, and each of those switches alters the reported percentage for the same document [3]. A 22% score with quotes and references excluded is a very different document from a 22% score with everything counted. Before reacting to a percentage, check which filters were applied — the report header states them.
A high score can be entirely legitimate, and a low score can still contain problems. A literature review that quotes heavily may legitimately sit in the twenties or thirties, while a 4% score can hide one unattributed paragraph that happens to be the most important one in the essay [3]. The score is a triage signal that tells you where to look; the highlighted matches tell you what is actually there. Treat the percentage as an index into the report, not as a summary of your integrity.
Can You Check Your Own Document Through a Similarity Detection System Before You Submit It?
Sometimes yes, and the answer depends entirely on your instructor's assignment settings. Many institutions enable draft or self-check submissions, where a student can run a paper through the system and see the report without that submission counting as the final one; other assignments deliberately disable this so that the first submission is the only one [4]. If the option is not visible in your course, it has not been switched on — there is no hidden student-facing portal that bypasses the instructor's configuration.
Where self-check is unavailable, students commonly use an independent pre-submission check to see the same style of report before the real deadline. The value is timing: a similarity report is most useful when you still have hours to fix an uncited paraphrase, add quotation marks, or reword a passage that matched a source you had genuinely never read [4]. Seeing the report after the deadline converts a fixable citation problem into an academic-integrity conversation.
The practical workflow that works is narrow and repeatable. Run the check, open the report, and work through the matches in descending order of size rather than staring at the headline percentage; fix the largest blocks first, because they carry the most weight in both the score and the reader's impression [4]. Then re-check once. Two passes catch almost everything a student can catch on their own, and the second pass confirms that your edits actually removed the overlap rather than just reshuffling it.
If you would rather see the actual report than guess at it, turnitin0 runs the same comparison and hands you the document you would otherwise only meet after the deadline — a real Turnitin AI report and a real similarity report, delivered together, so the matches you fix are the matches that matter.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is a similarity detection system the same thing as a plagiarism checker?
No. A similarity system reports text overlap with sources; plagiarism is a judgement about attribution and intent that a human makes after reading those matches [1]. The tool supplies evidence, the institution supplies the verdict.
Why is my similarity score higher than the percentages listed next to each source?
Because the score counts each matched phrase once, while a single phrase can be listed under several sources — so source percentages can add up to more than the headline figure [3]. This is expected behaviour, not an error in your report.
Does running a check through a similarity detection system add my paper to the student database?
Only if the check is configured that way. Repository inclusion is an account-level setting, and non-repository checks compare your document against the corpus without retaining it [2].
Can I see my similarity report before the assignment deadline?
Only if your instructor has enabled draft or self-check submissions on that assignment [4]. If the option is absent from your course, it is disabled rather than hidden.
Does a high similarity score automatically mean I will be penalised?
No. Scores rise with legitimate quoting, common phrasing, and reference lists, and report filters for quotes and bibliography change the number substantially [3]. What matters is whether each highlighted match is properly attributed.