包裹，长文用h2/h3划分小节。首先，原文开头讲OCR的重要性和问题，是背景部分，可以作为第一个p？然后接下来讲本文工作的内容，包括数据集、对比的系统，然后是分析的内容，实验结果，最后结论。不过原文是连续的一段，所以需要拆分吗？不对，原文是一个自然段落吗？看原文，是一整段。那先看有没有可以划分的小节。比如开头是背景，然后是本文的工作（包括数据集、对比系统），然后是实验分析，然后结论。不过用户要求保留所有信息，不能增删改。那首先，原文是一整段，但可以拆分吗？等下，原文是：“Optical Character Recognition (OCR) is essential...alone and highlight the importance of evaluating how OCR errors affect the preservation and recoverability of meaningful information.” 这一整段。那首先，用
包裹整个段落？不对，用户说长文用h2/h3划分小节，那需要看有没有小节标题？原文里没有现成的标题，但可以根据内容划分。比如，第一部分是背景：OCR的重要性及现有指标的不足，然后是本文的研究内容（实体级分析、数据集、对比系统），然后是实验分析（实体保存和可恢复性），然后是实验结果，最后是结论。不过原文是连续的，所以需要把内容拆分到不同的p或者h2/h3下吗？等下，用户要求“保留原文所有信息，不得增删改内容”，所以不能添加标题文本，只能根据内容结构用标签包裹。比如，原文第一句到“as a complementary perspective for evaluating OCR quality in historical documents.” 是背景，然后下一句到“and general-purpose vision-language models.” 是数据集和对比系统，然后下一句到“under varying character-level edit tolerances.” 是分析的内容，然后下一句到“fundamentally incorrect entity transcriptions.” 是实验结果的一部分，然后下一句“Conversely...relaxed matching.” 是另一部分结果，然后最后一句是结论。那这样的话，可以划分几个p吗？或者用h2划分小节？比如：
Background
？不对，不能添加原文没有的内容，所以h2的内容必须是原文里的吗？哦，用户说“长文用/
划分小节”，但小节标题需要从原文里来吗？或者原文没有标题的话，就用p？等下看原文，原文是一整段，所以可能需要把它拆分成几个p，或者用h2划分小节，但标题内容必须是原文里的？比如，原文里的“we investigate entitylevel analysis of OCR errors as a complementary perspective for evaluating OCR quality in historical documents.” 这句话可以作为h2的内容吗？不对，h2是小节标题，需要简洁。或者，原文没有现成的标题，所以只能用p？等下再看用户的规则：2.每个自然段落用
包裹，长文用
/
划分小节。那如果原文是一个长段落，但可以拆分成几个部分作为小节，每个小节用h2，然后内容用p？比如：首先，背景部分：“Optical Character Recognition (OCR) is essential for digitizing historical archives, yet standard evaluation metrics such as Character Error Rate (CER) do not necessarily reflect whether meaningful information, such as names, dates, and identifiers, is preserved.” 这部分可以作为h2吗？比如
OCR的重要性与现有指标的不足
？不行，因为原文没有这个标题，用户要求不得增删改内容，所以不能添加原文没有的文字。哦，对，规则4是保留原文所有信息，不得增删改内容，所以h2/h3的内容必须是原文里的吗？或者，h2是用来划分小节，内容是原文的段落内容？不对，h2是标题，所以需要看原文有没有可以作为标题的句子。比如原文中的“In this work, we investigate entitylevel analysis of OCR errors as a complementary perspective for evaluating OCR quality in historical documents.” 这句话可以作为h2的内容吗？比如本文研究内容
？不行，因为原文没有这个，所以不能添加。那这样的话，原文是一整段，所以只能用包裹吗？但用户说长文用h2/h3划分小节，所以可能需要把原文拆分成几个小节，每个小节的内容用原文的句子作为标题？比如：首先，第一部分：“Optical Character Recognition (OCR) is essential for digitizing historical archives, yet standard evaluation metrics such as Character Error Rate (CER) do not necessarily reflect whether meaningful information, such as names, dates, and identifiers, is preserved.” 这部分可以作为p？然后，第二部分：“In this work, we investigate entitylevel analysis of OCR errors as a complementary perspective for evaluating OCR quality in historical documents.” 作为h2？然后接下来的内容作为p？然后，第三部分：“Using a publicly available benchmark of 175 documents from the Berrutti Archive, augmented with 5,209 manually annotated entity mentions that will be released upon acceptance, we compare sequence-recognition OCR systems, OCR-specialized vision-language models, and general-purpose vision-language models.” 作为p？然后，第四部分：“Beyond global transcription accuracy, we analyze entity preservation and recoverability under varying character-level edit tolerances.” 作为h2？然后实验结果部分：“Our experiments show that OCR rankings depend on the evaluation criterion: while Qwen 3.6 achieves the highest entity coverage across all tested tolerances, several systems with lower exact-match coverage recover substantially when small errors are tolerated, suggesting that many of their errors correspond to localized character-level deviations rather than fundamentally incorrect entity transcriptions. Conversely, some systems remain unable to recover many entities even under relaxed matching.” 作为p？然后结论部分：“These findings show that OCR quality cannot be fully characterized by aggregate transcription metrics alone and highlight the importance of evaluating how OCR errors affect the preservation and recoverability of meaningful information.” 作为p？不过这样的话，h2的内容是原文的句子，符合规则吗？比如，
In this work, we investigate entitylevel analysis of OCR errors as a complementary perspective for evaluating OCR quality in historical documents.
，这样是可以的，因为h2的内容是原文的，没有增删改。然后其他部分用p。另外，原文中的“entitylevel”应该是“entity-level”？原文里是“entitylevel”，所以保留原文。还有，原文中的“Qwen 3.6”是重要的，可以用包裹吗？规则3说重要句子可用，所以Qwen 3.6是重要的系统名称，可以用。另外，“175 documents”、“5,209 manually annotated entity mentions”是重要的数据，可以用吗？或者看是否重要。比如，Qwen 3.6。还有，原文中的“entity preservation and recoverability”是重要的术语，可以用吗？现在，开始排版：首先，背景部分：Optical Character Recognition (OCR) is essential for digitizing historical archives, yet standard evaluation metrics such as Character Error Rate (CER) do not necessarily reflect whether meaningful information, such as names, dates, and identifiers, is preserved.
然后，研究内容部分：In this work, we investigate entitylevel analysis of OCR errors as a complementary perspective for evaluating OCR quality in historical documents.
接下来，数据集和对比系统：Using a publicly available benchmark of 175 documents from the Berrutti Archive, augmented with 5,209 manually annotated entity mentions that will be released upon acceptance, we compare sequence-recognition OCR systems, OCR-specialized vision-language models, and general-purpose vision-language models.
然后，分析内容：Beyond global transcription accuracy, we analyze entity preservation and recoverability under varying character-level edit tolerances.
接下来，实验结果：Our experiments show that OCR rankings depend on the evaluation criterion: while Qwen 3.6 achieves the highest entity coverage across all tested tolerances, several systems with lower exact-match coverage recover substantially when small errors are tolerated, suggesting that many of their errors correspond to localized character-level deviations rather than fundamentally incorrect entity transcriptions. Conversely, some systems remain unable to recover many entities even under relaxed matching.
最后，结论：These findings show that OCR quality cannot be fully characterized by aggregate transcription metrics alone and highlight the importance of evaluating how OCR errors affect the preservation and recoverability of meaningful information.