Direct answer
A plagiarism checker is software that compares a document against a defined set of source collections — a student paper repository, academic journals and publications, and indexed web pages — and returns a similarity score plus a list of matched sources [1]. The score is a measure of overlapping text, not a verdict on intent: a high score can come from correctly quoted material, and a low score does not prove original argument [1]. What separates one checker from another is almost entirely the database behind it.
What databases does a plagiarism checker search?
Turnitin's comparison set is the standard reference point for academic work because it combines three distinct layers: a student paper repository containing work previously submitted to institutions, a large body of proprietary and open-access academic journals and publications, and billions of current and archived web pages [2]. The repository layer is the one most students underestimate. A passage recycled from an essay submitted two years ago at a different university may not exist anywhere on the open web, yet it can still surface as a match if the institution's licence includes the repository [2].
Coverage is broad but never exhaustive. Turnitin is explicit that no checker indexes every book, every paywalled article, or anything in print that was never digitised, so a clean report means "no match found in these collections," not "no unattributed text exists" [2]. Institutional licensing also changes the answer: whether the student repository is included in a given check depends on the agreement the university has in place, which is why two students running the same file at different institutions can receive different match lists [2].
For practical purposes, treat database coverage as the first question you ask about any checker. A tool that searches only the open web will miss repository matches, translation plagiarism, and paywalled journal reuse — precisely the categories that matter most in coursework and dissertations [2]. A tool that searches a repository but no journal index will miss the opposite cases. The useful question is not "does it have a database" but "which layers does it have, and does that match what my department will run" [2].
Why do different plagiarism checkers give different similarity scores for the same text?
Three variables explain almost all score differences between tools: the underlying databases, the matching algorithm, and the settings applied to quotes, bibliography, and small matches [3]. If one checker holds a journal index and another does not, the same paragraph will match in one report and vanish in the other. If one tool excludes quoted material and the reference list from the total while another counts them, a well-cited essay can swing by many percentage points without a single word changing [3].
Algorithm behaviour matters too. Some tools merge several short matches from the same source into one entry, others list each fragment separately; some ignore matches under a threshold of a few words, others report them [3]. This is why a 34% score from one service and a 12% score from another is not evidence that either is broken — it is evidence that they are measuring different things against different collections [3].
The practical implication is to read the report, not the headline number. A single high match to one source is far more actionable than a diffuse total spread across forty small matches, because the single match tells you exactly which passage to paraphrase or cite properly [3]. When you compare tools, compare the source list: the names, the match percentages per source, and whether the highlighted text is genuinely reused or just a common phrase [3].
How can a student preview the same similarity report their university will see?
Students benefit most from seeing a similarity report while there is still time to act on it. A pre-submission view lets you fix paraphrasing that drifted too close to the original, add a missing citation, or restructure a paragraph that leaned on one source — changes that are impossible once the deadline passes [4]. Because the report highlights matched text against each named source, revision becomes targeted rather than guesswork [4].
The critical detail is report fidelity. A preview is only useful if it reflects the same comparison set and the same scoring behaviour as the institutional check; a web-only scan gives false comfort precisely where repository and journal matches would have appeared [4]. Reading the report also requires calibration: the goal is correct attribution, not a zero score, and a report with a modest percentage and clearly cited sources is usually healthier than a suspiciously empty one [4].
Turnitin0 exists for that gap. You upload a .docx, .pdf, or .txt file (English, over 300 and under 30,000 words, under 20 MB) and receive two PDFs in one checkout — a Turnitin AI detection report and a similarity report matching what instructors see in their LMS. Files are checked without being added to the student paper database, reports are not shared with third-party databases, and turnaround is under 15 minutes in 98% of cases. Turnitin0 is an independent service and is not affiliated with Turnitin, LLC.
If your draft is ready and you want to see the source list before your department does, turnitin0 gives you the same report anatomy — cover, score, flagged passages, and a similarity summary broken down by source — so you can revise with evidence instead of hoping.
※ 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 score is the percentage of text matching sources in the database, and correctly quoted or properly cited material contributes to it [1]. Instructors interpret the score alongside the source list and the assignment context.
Can a plagiarism checker find work submitted at another university?
Sometimes. Matches from previously submitted papers depend on whether the student paper repository is included in the institution's licence, so results vary between institutions [2].
Why did my score drop after I excluded quotes and references?
Because those elements were being counted in the original total. Excluding quotes, bibliography, and small matches is a standard setting that can move a score by many percentage points with no change to the text [3].
Does a clean report guarantee I am safe?
No. No database is exhaustive, so a low score means no match was found in those collections — not that every source has been accounted for [4].
Should I aim for a 0% similarity score?
No. A modest score with clearly attributed sources is normal and often healthier than an artificially empty report; the objective is correct attribution, not a zero [4].