Note: The thresholds and tools mentioned in this article reflect publicly available information as of April 2026. Always verify the specific regulations of your faculty, as these may differ.

What a plagiarism check is

If you are submitting a bachelor’s thesis in Romania, you have certainly encountered the term “plagiarism check” (or, in Romanian, verificare antiplagiat). Many students associate it automatically with the idea of intellectual theft, but the reality is more nuanced.

A plagiarism check is, in essence, a similarity scan: an algorithm compares your text against millions of indexed sources (academic articles, digitised books, websites, databases of previous theses) and reports the percentage of passages that also appear elsewhere.

The system cannot determine intent. It does not know whether you copied or quoted correctly, whether you are repeating a standard phrase from the field or whether you reproduced a paragraph without citing the source. Its role is to flag; interpretation belongs to the examination committee.

A similarity percentage is therefore not an automatic verdict; it is the starting point for deeper analysis. A report that you understand properly shows you exactly what needs correcting, allowing you to submit a clean document without unnecessary anxiety before your defence.

How the verification system works

Most public universities in Romania use Sistemul Antiplagiat (sistemantiplagiat.ro), a national platform integrated into the administrative workflow of higher education institutions. Theses are uploaded by the supervisor or directly by the student, and the report is generated automatically and added to the graduation file.

Beyond the national platform, some faculties with an international profile or external partnerships (particularly master’s and doctoral programmes) use Turnitin, a global platform with a broader database of international sources. The underlying principle is the same: your text is broken into fragments, each fragment is compared against indexed sources, and matches are flagged and aggregated into an overall percentage.

Both systems allow certain categories of text to be filtered out: the bibliography, quotations in standard format, and passages below a certain word count. If these filters are not correctly activated by administrators, the report may appear more “problematic” than it actually is; one more reason to understand what you are reading, not just look at the final number.

What percentages are accepted

The question “what similarity percentage is acceptable for a bachelor’s thesis?” has no single answer, but there is a clear pattern: most Romanian universities accept a similarity threshold of between 10% and 20%.

Some reference points:

  • Below 10%: generally unproblematic. However, a score of 0% is almost impossible to achieve legitimately and may itself raise questions. Standard field terminology, established definitions, and correctly marked quotations all contribute naturally to similarity.
  • 10–20%: the range in which the large majority of properly written theses fall. Committees assess whether the percentage derives from citations and common phrasing or from uncited reproductions.
  • 20–30%: a zone requiring attention. The report will be examined more carefully; if the similarity originates from correctly cited sources, the situation remains manageable.
  • Above 30%: a genuine risk of rejection, depending on the faculty’s regulations. Substantial revisions are necessary.

The exact threshold varies: some technical faculties accept up to 25% (due to the large number of formulae, definitions, and standards), while humanities faculties may apply stricter limits. Check your faculty’s internal regulations: that is the only document that matters before the committee.

How to read the report

The system-generated report contains more than a single number. Your text appears colour-coded: each highlighted passage indicates a match with an external source.

What to look for:

  • Colour/intensity of highlighting: passages with exact matches are marked more prominently than those with partial similarity.
  • The source of the match: the report shows you where the similar passage originates: an article, a previous thesis, a website. If the source is correctly cited in your text, the flag is not a problem.
  • Length of the matched passage: a match of 3–4 words on a common expression (e.g. “in conclusion, it can be stated that”) is irrelevant. A match of 20+ words within a descriptive paragraph is a clear signal.

What you can legitimately disregard:

  • Standard titles and headings
  • Mathematical and chemical formulae
  • Quotations explicitly marked with quotation marks and a full reference
  • Technical domain terminology that has no reasonable alternative

What to do if the threshold is exceeded

The first instinct (panic) is not helpful. A report with a high percentage tells you where, not necessarily what is wrong. Work through the steps below methodically.

1. Identify the origin of the similarity. Open the report and group the flagged passages. Do they originate from a single source or from several? If 80% of the similarity comes from a single thesis or book, the problem is localised and solvable.

2. Check whether your citations are correct. Many high percentages stem not from intent but from an incomplete bibliography. If you paraphrased an author without citing them, adding the reference in the text can resolve a significant portion of the problem. See the bibliography guide for correct source formatting according to UBB, ASE, and UB standards.

3. Check whether your paraphrase is genuine. A common mistake: you believe you paraphrased, but you only replaced a few words with synonyms. The system detects sentence structure, not just vocabulary. (See the next section for concrete examples.)

4. Check whether you reused your own prior texts without self-citation. If you published an article or submitted a previous paper on the same topic without citing it, you generate similarity against your own text.

5. Prioritise long passages. Do not spend time on 3–5 word matches. Focus on blocks of 20+ words: these raise the percentage the most.

How to paraphrase correctly (genuine paraphrase)

Real paraphrase means taking the idea, not the structure. Here are concrete examples of the difference:

Example 1

Original source: “Chronic stress affects the immune system and increases susceptibility to illness.”

Weak paraphrase (detectable): “Long-term stress influences immunity and raises vulnerability to disease.” Note: The structure is identical; only synonyms have been substituted.

Correct paraphrase: “Prolonged exposure to stressors diminishes the body’s capacity to defend against pathogens, according to research in psychoneuroimmunology (Author, Year).” Note: The idea is rendered from a different perspective, with added context and a citation.

Example 2

Original source: “Digitalisation of public administration reduces operational costs and improves citizens’ access to services.”

Weak paraphrase: “The digital transformation of administration lowers expenditure and facilitates public access to services.” Note: Nearly identical in structure.

Correct paraphrase: “The adoption of digital solutions in the public sector generates significant budget savings and simplifies the interaction between institutions and citizens, removing some of the traditional logistical barriers (Author, Year).” Note: Reformulated perspective, new details, citation present.

Example 3

Original source: “Intrinsic motivation produces more durable results than extrinsic motivation in educational contexts.”

Weak paraphrase: “Motivation that comes from within yields more stable results than motivation from outside in education.” Note: Synonymic translation, not paraphrase.

Correct paraphrase: “Students who study out of curiosity or genuine interest demonstrate better long-term knowledge retention than those motivated exclusively by external rewards or sanctions (Author, Year).” Note: The idea is developed with examples, not merely reformulated lexically.

The simple rule: if you can recognise the structure of the original sentence in your version, you have not paraphrased sufficiently.

Checking yourself before submission

The best time to discover a problem is before submission, not on the day you hand in the file. An early check gives you time to correct, reformulate, and adjust the bibliography without the pressure of a deadline.

Holipore offers a plagiarism check and technical assistance service through which you receive the similarity report together with an analysis of high-risk passages. We do not write or reformulate your text; we provide the tools and explanations you need to make the adjustments yourself, with a real understanding of the issue.

Mistakes that lead to high percentages

  • Copy-paste without citation: the most frequent case; text reproduced in full, without quotation marks and without a reference.
  • Superficial paraphrasing: synonyms instead of genuine reformulation of the idea.
  • Over-reliance on a single source: one author or article cited ten times generates a high percentage even when cited correctly.
  • Reusing your own previous texts without self-citation: prior essays, papers, or projects are indexed; self-citation is obligatory.
  • Bibliography copied from another thesis: a reference list reproduced as-is raises the percentage.
  • Generic boilerplate text: introductions or conclusions worded in a “standard” way without adaptation are detected as similar to other theses in the field.

Myths about plagiarism checks

Myth 1: “If I change the synonyms, the system won’t detect it.” False. Modern algorithms analyse syntactic structure, not just vocabulary. Replacing key words with synonyms does not evade detection.

Myth 2: “A score of 0% means the thesis is perfect.” False. A score of 0% is practically impossible for a properly documented academic thesis. It may indicate that the text could not be compared correctly, or that no indexed sources exist; not that the thesis is “better”.

Myth 3: “If I translate from a source in another language, it won’t be detected.” Partially true in the past; increasingly less so now. Turnitin and other advanced platforms have cross-linguistic detection modules. In any case, translating without citing remains plagiarism.

Myth 4: “If I embed the text as an image in the PDF, the system won’t read it.” False for systems that apply OCR (optical character recognition). And even if it worked technically, the examination committee reading the thesis would notice the visual anomaly.

Conclusion: when to seek assistance

A plagiarism check is not an arbitrary obstacle; it is an instrument of academic integrity. A “problematic” report shows you precisely where you need to do more work: on citation, on paraphrasing, on diversifying your sources.

If the threshold is exceeded, approach it systematically: locate the high-risk passages, verify that citations are complete, reformulate genuinely (not cosmetically) and run a second check before submission.

If you want to remove uncertainty and submit with the report already reviewed, our plagiarism check service provides both the technical report and editorial guidance for interpreting and addressing the flagged passages. You remain the author of your thesis; we provide the technical and methodological support to ensure it meets academic standards.