通过八项要求降低文本的AI痕迹:提高句式与词汇多样性、适当增加困惑度与突发性、减少机械连接词、增加语义复杂性、调整词汇频率分布与可读性水平,避免高频AI用语,让改写后的文本更贴近真人自然写作风格。
中文版提示词
请对以下原文进行润色,目标是最大限度降低AI痕迹: 1. 提高句式多样性:交替使用长句和短句,增加主动句、被动句、疑问句和感叹句,避免过于规则单调的句式。 2. 优化词汇多样性:避免频繁重复相同短语或词汇,替换常见高频用词,适度引入同义词和近义表达。 3. 提高困惑度:适当引入少见词汇或短语,避免过于常规、可预测性强的表达,可适度使用具有个人风格或专业领域独有的术语。 4. 增强突发性:刻意在文本中创造语气和语调上的变化,不保持过于一致平稳的风格。 5. 减少连接词的机械使用:避免机械使用"首先""其次""再次""然而"等连接词,使逻辑连接更自然灵活。 6. 增加语义复杂性:适当使用隐喻、类比和修辞手法,使文本更贴近人类自然写作习惯。 7. 调整词汇频率分布:避免反复使用相同短语或习惯用语,尤其是典型AI偏好用语(如"深入探讨""不容忽视"等)。 8. 调整可读性水平:避免内容过于简单或复杂,保持适中可读性,穿插使用简单与复杂句式。 以下是我需要润色的原文:
英文版提示词
Please polish the following original text with the goal of minimizing AI traces as much as possible: 1. Increase sentence variety: alternate long and short sentences, and add active, passive, interrogative, and exclamatory sentences to avoid overly regular and monotonous patterns. 2. Optimize vocabulary variety: avoid frequently repeating the same phrases or words, replace common high-frequency terms, and moderately introduce synonyms and near-synonyms. 3. Increase perplexity: appropriately introduce uncommon words or phrases, avoid overly conventional and predictable expressions, and use terms with personal style or field-specific flavor where suitable. 4. Enhance burstiness: deliberately create changes in tone and register, avoiding an overly consistent, flat style. 5. Reduce mechanical use of connectives: avoid mechanical use of words like "first," "second," "next," "however," and make logical connections more natural and flexible. 6. Increase semantic complexity: appropriately use metaphor, analogy, and rhetorical devices to make the text closer to natural human writing. 7. Adjust word frequency distribution: avoid repeatedly using the same phrases or habitual expressions, especially typical AI-favored phrases (e.g., "delve deeply," "cannot be overlooked"). 8. Adjust readability level: avoid content that is too simple or too complex; keep readability moderate and intersperse simple and complex sentence structures. Here is the original text I need polished: 🛠️ **适用 AI 工具**:ChatGPT、Claude、Kimi、豆包、DeepSeek、文心一言

◯ 评论 0