Plagiarism Detection Systems: A Comparative Guide to Domestic and International Tools and Rewriting Strategies

Key Points

How Major Detection Systems Work

Chinese Domestic Systems:

• CNKI (知网): Uses continuous character matching with fuzzy recognition, flagging matches of 13 or more consecutive characters with a 5% paragraph threshold. Includes a unique "Undergraduate Thesis Joint Comparison Library." Over 90% of Chinese universities use it as the final-check system.
• VIP (维普): Employs semantic text segmentation that can identify paraphrased ("pseudo-original") content. Strong in science and technology fields. Requires similarity below 20%; content above 30% may be rejected outright.
• Wanfang (万方): Uses a sentence-level orthogonal soft clustering inverted semantic algorithm, delivering minute-level results at lower cost. Strong authority in medical literature.

International Systems:

• Turnitin: Uses NLP and deep semantic analysis to detect "smart plagiarism" and cross-language plagiarism. Database covers 56,000 journals, 90 billion web pages, and 1 billion student papers.
• iThenticate: Turnitin's sister product for academic publishers, additionally including ProQuest dissertations. Used by Nature, Elsevier, and other top-tier journals.

Core Differences: Form Matching vs. Content Matching

| Dimension | Chinese Systems | International Systems | |-----------|----------------|---------------------| | Algorithm | Form-focused character matching | Content-focused NLP semantic analysis | | Database | Chinese literature, dissertations | Global multilingual resources | | Citation Recognition | Weaker; depends on strict format | Smarter contextual understanding | | Report Interpretation | Total similarity ratio as pass/fail metric | Analyzes content nature and source | | Strictness | CNKI moderate, VIP strict, Wanfang lenient | Generally strict; detects cross-language plagiarism | | Use Cases | Chinese university theses | International publication, cross-border research |

Rewriting Strategies

For University Theses:

• Standard approaches: Follow GB/T 7714 citation format; paraphrase in your own words while restructuring sentences and argumentation angles
• Tactical approaches: Avoid long direct quotations; convert data descriptions into charts and tables, which detection algorithms handle less effectively
• Process approaches: Run iterative check-modify-recheck cycles using the same system as your institution

Golden rules for thesis rewriting:

• Start checking early to allow sufficient revision time
• Use the same detection system as your institution for pre-checking
• Prioritize chapters with highest similarity rates
• Maintain logical clarity without sacrificing quality

For Journal Papers:

• Enhance originality and research depth: Identify gaps in existing research, contribute unique theoretical frameworks, and demonstrate novel problem formulations
• The fundamental approach is genuine research innovation rather than surface-level text manipulation

Conclusion

International systems like Turnitin offer superior global coverage with NLP-based semantic analysis, while domestic systems excel at Chinese-language authority and formal matching. For thesis writers, understanding the specific algorithm of their university's chosen system is critical for effective rewriting. For journal publication, the highest-value strategy remains genuine research contribution rather than technical evasion of detection tools.