Turnitin and iThenticate reduce AI strategy, and AI rate drops below 20%

Many authors have their papers rejected due to excessively high similarity rates when submitting. This article will elaborate on the standards of major plagiarism check systems and provide practical techniques to reduce similarity rates, helping you easily lower the similarity rate to below 20%.

1 Differences in Plagiarism Check Standards

Hello everyone, the editor is online again to share knowledge about academic papers with you! Many authors struggle with similarity rate issues when submitting their papers. Clearly, the content is original, so why is the similarity rate still high? Today, the editor will teach you how to reduce the similarity rate of your paper and guide you step by step to lower it to below 20%!

The similarity rate requirements of Turnitin and iThenticate differ from each other:

• Turnitin has stricter requirements for journal papers, requiring the similarity rate to be controlled between 15% and 20%.
• iThenticate, however, has relatively looser standards, with the similarity rate generally allowed to range from 20% to 30%.

Understanding these differences is crucial. In practice, a common scenario occurs: even if authors adhere to academic norms and avoid plagiarism, the final similarity rate may still be high. From this, we can draw a conclusion: a high similarity rate ≠ plagiarism.

The standards of different plagiarism check systems vary significantly:

Turnitin applies strict criteria: it marks text in red when 13 consecutive characters are duplicated, and it also compares the paper against the "Academic Paper Joint Comparison Database".

iThenticate is also quite strict: even function words like "is " and "so on" are counted as duplicates, and the scope of content marked in red is often expanded.

2 Causes of High Similarity Rates and AI-Generated Content Rates in Papers

Several factors contribute to a high similarity rate, including: directly copying content for literature reviews from textbooks or encyclopedias; plagiarizing others’ experimental data when describing data; and copying the derivation process of classical theories when deriving formulas.

Specifically, the high similarity rate of papers mainly stems from the following three points:

Excessive citations in the literature review section;

High similarity in the description of experimental methods;

Lack of originality in data interpretation.

Therefore, when facing a high similarity rate (or AI-generated content rate) in a paper, authors do not need to panic excessively, nor should they easily equate it with academic misconduct. Instead, they should first understand how plagiarism check systems work, then conduct "academic coding" and "content reshaping" on the text in a targeted manner. It should be noted that this is not to encourage authors to deliberately evade checks through technical means, but to help them write papers more standardizedly and with greater originality. After understanding the root causes of the problem, authors can proactively adopt the following effective strategies to optimize their manuscripts.

3  How to Reduce Similarity Rates and AI-Generated Content Rates

Many authors feel confused and find it difficult to reduce similarity rates (and AI-generated content rates), but in fact, it is not that hard. Here are several practical techniques:

Paraphrasing: Replace "direct copying" with "paraphrasing". Avoid copying the original text directly; flexible use of paraphrasing skills is key.

Sentence Expansion: Expand short sentences into more detailed long ones.

Reduction: Refine "redundant content" to retain the "core essence".

Visualization (Chart Method): Convert "textual information" into "visual images" (e.g., charts, graphs).

Citation Method: When citing sources, be sure to mark the reference (e.g., [1]) and add your own insights. For example: "As Zhang San (2020) stated... This view provides theoretical support for this paper, and this paper further conducts in-depth research based on this perspective...". This can effectively reduce the text’s similarity level (and traceability of AI-generated content).

While mastering these methods is important, it is even more critical to understand that the essence of these techniques is "content reshaping" rather than "simple replacement". Their purpose is to improve the expression of your paper, not just to evade plagiarism checks. If you mechanically apply these methods without considering the coherence of meaning and the logic of the paper, you will fall into a new predicament—the "similarity reduction trap".

Be cautious of the following three types of "similarity reduction traps":

Replacing words only based on literal meaning (e.g., replacing "research" with "explore" and "method" with "the way of"). This easily leads to unsmooth sentences and even distorts the original meaning.

Simply adjusting the order of sentences (e.g., changing "A leads to B, and B affects C" to "B affects C because A leads to B"). This modification may still be detected as duplication by plagiarism check software.

Over-relying on similarity reduction tools. Some of these tools may generate sentences with ambiguous meanings or grammatical errors.

4 Conclusion

In essence, reducing similarity rates (and AI-generated content rates) means "rephrasing content in your own words on the basis of fully understanding the original meaning", rather than simply replacing synonyms or adjusting sentence order.

It should be noted that plagiarism checks usually require payment. Therefore, everyone should attach importance to the plagiarism check process when writing papers. It is recommended to first use free plagiarism check websites (such as GPTZero and ZeroGPT) for preliminary detection. After confirming there are no obvious issues, use paid plagiarism check websites for accurate testing.

If the similarity rate is too high, the parts marked in red in the paper are the key areas that need revision. You only need to modify these red-marked parts and then conduct a recheck after revision!

That’s all for Author’s sharing today! The first step to publishing a paper is to pass the plagiarism check. Follow the editor, and the editor will bring you more knowledge about academic papers!