I used these methods to reduce weight and AI, and the paper repetition rate was reduced from 55% to 6%
Reducing Similarity Rate Is Not a "Word Game". On the premise of not undermining academic rigor, reducing similarity rate involves reorganizing the text logically and optimizing its expression, so that the textâs similarity rate meets the requirements (undergraduate papers: < 30%, masterâs and doctoral dissertations: < 15%). This article will discuss the topic from three aspects: analysis of text similarity rate, text rewriting strategies, and compliance detection.
1. Causes of High Similarity Rate
1.1 Classification of Duplication Types
Duplication can be classified into the following types:
- Direct copying (plagiarizing the original text without marking citations);
- Similar expression (identical core viewpoints with similar sentence structures);
- Data and chart duplication (directly using othersâ charts without redrawing or marking the source).
- With the help of the "full-text red-marking" function inTurnitinâs similarity check report, duplicate content can be categorized and counted based on characteristics such as "duplicate segment length > 100 characters" and "duplicate sources (journals, dissertations)". Priority should be given to addressing parts with high similarity rates (e.g., over 50%).
1.2 Duplication in Literature Citations
Principles for identifying valid citations: Distinguish between "reasonable citations" and "excessive citations". When citing othersâ viewpoints, use the format "Author + Year + Page Number" (e.g., Li Si, 2022: 45). Additionally, cited content should not account for more than 30% of the corresponding paragraph.
For red-marked content that belongs to "common knowledge" (such as basic disciplinary definitions), it is necessary to reorganize it using "personalized expression" to avoid duplication with existing literature.
2. Strategies for Rewriting to Reduce Similarity Rate
2.1 Optimization of Sentence Structures and Vocabulary
Do not rigidly adhere to the "subject-verb-object" structure. Instead:
- Convert passive voice to active voice (e.g., change "The study found XX" to "XX has been confirmed by the study");
- Split long sentences into short ones (e.g., split sentences of over 50 characters into 2-3 shorter sentences);
- Replace "high-frequency words" with logical conjunctions or alternative terms: use conjunctions like "Furthermore" or "However" appropriately; replace "influence" with "mechanism of action" or "effect"; replace "analyze" with "dissect" or "explore".
- It is important to ensure the accuracy of academic termsâfor example, "reliability and validity" must not be arbitrarily altered.
2.2 Reconstruction of Paragraph Logic
For duplicate paragraphs:
- Extract the "core viewpoints" (e.g., a theory includes three dimensions: A, B, and C);
- Rearrange them in a "new logical order" (e.g., change from "dimension definition â dimension relationship â application scenario" to "application scenario â dimension relationship â dimension definition");
- Add "personal interpretations": After citing othersâ viewpoints, supplement with content like "Combined with the research theme of this article, this viewpoint can be further expanded to...", which helps enhance the uniqueness of the content.
3. Compliance Check and Recheck for Similarity Rate
3.1 Recheck of Similarity Rate
Use the similarity check systems designated by the university (such as Turnitin and iThenticate) for rechecking. If the similarity rate still exceeds the limit, focus on revising "yellow-marked paragraphs" (mild duplication) by "adjusting sentence order" or "adding cases". Avoid using "free similarity check tools", as their data comparison databases are incomplete and the results are inaccurate.
3.2 Review of Citation Formats
Verify citation formats in accordance with journal or university regulations (e.g., the GB/T 7714 standard) to ensure consistency between "citation marks in the main text" and the "reference list at the end of the article". Confirm that information such as "author name, publication year, article title, and journal name" is accurate and error-free.
For charts, redraw them using software such as Origin or Excel, and add "figure/table number + title + data source" (e.g., Table 2: Descriptive Statistics of Variables (Data Source: Compiled in this article)) to avoid copyright issues.
Humanize AI text Without Login
Make your AI text 100% undetectable with HumanPass.pro - your trusted AI detection remover to create plagiarism-free, human-like text.