Originality Report

An originality report — also called a similarity report — is the automated output of a text-matching system that compares a submitted manuscript, thesis, or dissertation against a large reference corpus (previously published literature, archived and current web pages, and a repository of prior student submissions) and returns a similarity index: the percentage of the submitted text that matches content already in that corpus. The two dominant services in research administration are Turnitin Similarity, widely used by universities to screen coursework and electronic thesis/dissertation (ETD) deposits, and Crossref’s Similarity Check, a service publishers and journal editors use that is powered by Turnitin’s iThenticate engine and draws on Crossref members’ published content in addition to the same web and archival sources.

How the similarity index is calculated

The system segments submitted text into phrases, discards common “stop words,” and checks each phrase against the reference database, then aggregates the proportion of the document with a detectable match into a single percentage. A high similarity index does not distinguish, on its own, between properly quoted and cited material, a shared bibliography or reference list, standard methodological or boilerplate phrasing common to a field, and genuinely uncredited copying — a human reviewer has to make that distinction by opening the report and inspecting which passages are flagged and why.

How it is used in research administration

• Thesis and dissertation submission: many graduate schools require an originality/similarity report as a condition of electronic thesis or dissertation deposit, reviewed by the supervisor or graduate office alongside the candidate’s own declaration of originality.
• Journal submission pre-checks: editors commonly run a similarity check (frequently via Crossref’s Similarity Check / iThenticate) on newly submitted manuscripts before or during peer review, using the score as one screening input rather than an automatic accept/reject criterion.
• Research-integrity screening: an unusually high or unexplained similarity score can be one trigger, among several, for a closer editorial or institutional look — but per COPE guidance, any resulting action still follows a case-by-case investigation, not the percentage alone.

Why a similarity score is not a verdict

Turnitin’s own guidance is explicit that its Similarity tool “does not determine whether plagiarism has occurred” — it is designed to support a reviewer’s judgment, not replace it. A high score is routinely inflated by properly quoted and cited passages, a shared reference list or bibliography, standard field terminology, and — for a thesis — the candidate’s own previously deposited coursework or preprints (a case usually needing a specific self-matching exclusion, not treatment as third-party plagiarism). Conversely, a low or zero similarity score does not certify a work as free of misconduct: text-matching tools detect verbatim or lightly reworded textual overlap only, so they cannot themselves catch idea plagiarism (uncredited use of someone else’s concept, expressed in genuinely original wording), fabricated or falsified data, or self-plagiarism disclosed honestly but improperly cited. Reviewers should also expect the score to rise for text deliberately reworded through automated paraphrasing — see tortured phrases as the associated forensic signature — since crude synonym substitution can lower a similarity score without making the underlying appropriation any less improper.

References

• Turnitin, “Understanding the similarity score” (Turnitin Guides)
• Crossref, “Similarity Check” service documentation
• Committee on Publication Ethics (COPE), Core Practices and case guidance on plagiarism and text overlap