Direct answer
Direct Answer - ** Plagiarism detection is the process of comparing a submitted document against a large indexed corpus of web pages, academic journals, and previously submitted student papers to find passages that match existing sources [1]. The software returns a similarity report** that highlights each matched passage and links it to its origin, rather than issuing a verdict of guilt [1]. In practice, detection answers one narrow question — "does this text already exist somewhere else?" — and leaves the judgement about attribution, intent, and academic misconduct to a human reader [1].
How Does Plagiarism Detection Software Actually Work?
Detection begins with indexing. Systems such as Turnitin maintain a repository built from crawled web content, publisher databases, and the millions of student papers previously submitted to the service, and every new document is measured against that corpus rather than against a single "original" file [2]. The engine breaks the submission into text segments, then looks for segments that appear in the repository, which is why it can surface a match to a decade-old essay as easily as to a page published yesterday [2].
The headline number a student sees is the similarity index — the percentage of the paper's text that matched something in the repository [2]. That figure is a proportion of matched words, not a probability that the student cheated, and Turnitin's own guidance is explicit that a high index can be produced entirely by correctly quoted and cited material [2]. Many institutional configurations also exclude the bibliography, quoted passages, and matches below a small word threshold, so the same paper can show two different percentages depending on the settings an instructor applies [2].
It is worth separating plagiarism detection from AI writing detection, because the two are often confused. Similarity matching compares your text to existing text; AI detection models estimate whether a passage was generated by a large language model, and the two signals appear as separate reports [2]. A paper can score 0% similarity and still carry a high AI writing indicator, and vice versa [2].
What Does a Plagiarism Detection Report Show You?
The report is a reading tool, not a scoreboard. Each matched passage is colour-coded and numbered, and clicking a match opens a side-by-side view of the student's text next to the source it resembles, so a reader can judge whether the wording is properly attributed [3]. Because the match is shown in context, an instructor can distinguish a legitimate quotation with a citation from an unattributed paraphrase in seconds [3].
Reports are also configurable, and that configurability changes what you see. An instructor can switch off quotes, bibliography, and small matches to isolate the passages that actually need scrutiny, which is why two people looking at the same submission may quote different percentages [3]. The similarity layer sits alongside a separate AI writing indicator, and neither layer is designed to stand alone as a grade [3].
The practical takeaway for a student is that a report is diagnostic. It tells you where your writing overlaps with existing sources so you can add quotation marks, tighten a paraphrase, or insert a citation — the kind of revision that strengthens a paper rather than simply lowering a number [3].
Can Students Check Their Own Work for Plagiarism Before Submitting?
Usually not through the institutional route. When a class assignment is configured as a Turnitin dropbox, students normally cannot submit a draft and retrieve a report on demand, because the submission is the graded submission [4]. Depending on the instructor's repository settings, anything uploaded to a class dropbox may also be retained in the student paper database, which means a draft check can create its own match later [4].
That repository behaviour is the reason non-repository pre-submission checks exist. An independent check can return a downloadable similarity report and AI writing report without adding the file to the student paper database, so the draft does not become a future source for anyone else [4]. Turnitin0, for example, runs English documents between 300 and 30,000 words through a non-repository check and returns both reports together, with most orders completing in under 15 minutes.
Reading your own report before the deadline is the single most useful habit here. It converts an abstract definition of "plagiarism detection" into a concrete list of passages you can fix while you still have time to fix them [4].
If you have just learned what detection is, the fastest way to make it useful is to see a real report on your own draft — turnitin0 gives you the same AI and similarity reports your professor sees, before you submit.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Does a high similarity score mean I plagiarised?
No. The similarity index is the share of your text that matched something in the repository, and correctly quoted, properly cited material counts toward it [2]. A reader has to look at the highlighted passages to judge whether attribution is correct.
Is plagiarism detection the same as AI detection?
They are separate checks. Similarity detection compares your writing to existing published and submitted text, while AI detection estimates whether a passage was machine-generated [2]. Turnitin reports them as distinct signals.
Why did my instructor and I see different percentages?
Report settings differ. Quotes, bibliography, and small matches can be excluded, and each change alters the index without altering the paper [3].
Can I get a plagiarism report without my draft entering the student paper database?
Yes, through a non-repository pre-submission check. Turnitin0 checks your file without adding it to Turnitin's student paper database and returns the AI and similarity reports together, so the draft does not become a future source for other students [4].
What file types and lengths can be checked?
Turnitin0 accepts English .docx, .pdf, and .txt files between 300 and 30,000 words, under 20 MB, and delivers results in under 15 minutes in most cases.