首页 > AI前沿 > Learning Word-Level Explanations of Text Reuse in Latin Literature with Little Annotation

Learning Word-Level Explanations of Text Reuse in Latin Literature with Little Annotation

arXiv自然语言 2026-11-13 23:33 10 阅读 查看原文
用户需要我扮演一个专业的新闻排版工程师,将给定的一段纯文本新闻正文排版成语义化的HTML片段。 规则如下: 1. 只输出HTML片段本身,不要Markdown围栏,不要解释。 2. 每个自然段落用`

`包裹,长文用`

`/`

`划分小节。 3. 列表内容用`
  • `,引用用`
    `,代码用`
    `,重要句子可用``。
    4. 保留原文所有信息,不得增删改内容。
    5. 不要输出``/``/``等文档级标签。
    
    正文内容:
    Detecting an intertextual reference is only the first step; philologists also need to know how a later author reworked the source. We explain a reuse pair word by word: each word of the reusing passage is first aligned to its source word, then labeled with the operation that transformed it, such as a copy, a change of inflection, or a substitution. Such word-level operations are expensive to annotate, and only a few corpora annotated at this level exist for text reuse in Latin literature. We describe a procedure that needs only a limited number of annotations: synthetic pairs, built by applying known operations to Latin text, train a first model, and active learning chooses the pairs an annotator corrects. Among several architectures, a model that links and labels together performs best: a pointer network that, for every reuse word, chooses a source word and names the operation in the same step. We show that with active learning, 200 pairs chosen by uncertainty are enough for the model to match one trained on every annotated pair of the 1,490 references of the Loci Similes benchmark. We release our code, word-level predictions for all references, and an interactive demo.
    
    分析正文结构:
    这是一段学术新闻/摘要,只有一个自然段落。没有明显的列表、引用或代码。
    但是规则要求“每个自然段落用`

    `包裹,长文用`

    `/`

    `划分小节”。 由于原文只有一个自然段,我可以把它分成几个小节,或者只加一个`

    `。为了体现排版能力,我可以将这段长文按语义拆分成几个逻辑部分,并用`

    `划分小节,或者直接用`

    `包裹整个段落。 原文:“Detecting an intertextual reference is only the first step; philologists also need to know how a later author reworked the source. We explain a reuse pair word by word: each word of the reusing passage is first aligned to its source word, then labeled with the operation that transformed it, such as a copy, a change of inflection, or a substitution. Such word-level operations are expensive to annotate, and only a few corpora annotated at this level exist for text reuse in Latin literature. We describe a procedure that needs only a limited number of annotations: synthetic pairs, built by applying known operations to Latin text, train a first model, and active learning chooses the pairs an annotator corrects. Among several architectures, a model that links and labels together performs best: a pointer network that, for every reuse word, chooses a source word and names the operation in the same step. We show that with active learning, 200 pairs chosen by uncertainty are enough for the model to match one trained on every annotated pair of the 1,490 references of the Loci Similes benchmark. We release our code, word-level predictions for all references, and an interactive demo.” 可以拆分成:

    引言与词级对齐

    Detecting an intertextual reference is only the first step; philologists also need to know how a later author reworked the source. We explain a reuse pair word by word: each word of the reusing passage is first aligned to its source word, then labeled with the operation that transformed it, such as a copy, a change of inflection, or a substitution.

    标注挑战与流程

    Such word-level operations are expensive to annotate, and only a few corpora annotated at this level exist for text reuse in Latin literature. We describe a procedure that needs only a limited number of annotations: synthetic pairs, built by applying known operations to Latin text, train a first model, and active learning chooses the pairs an annotator corrects.

    模型架构与结果

    Among several architectures, a model that links and labels together performs best: a pointer network that, for every reuse word, chooses a source word and names the operation in the same step. We show that with active learning, 200 pairs chosen by uncertainty are enough for the model to match one trained on every annotated pair of the 1,490 references of the Loci Similes benchmark.

    资源发布

    We release our code, word-level predictions for all references, and an interactive demo.

    检查规则: 1. 只输出HTML片段本身,不要Markdown围栏、不要解释。 -> 是 2. 每个自然段落用`

    `包裹,长文用`

    `/`

    `划分小节。 -> 是 3. 列表、引用、代码等(此处没有,不需要强行加,但重要句子用了``)。 -> 是 4. 保留原文所有信息,不得增删改内容。 -> 检查拆分后的内容,文字完全一致,没有增删改。 5. 不要输出文档级标签。 -> 是 微调一下,确保文字完全一字不差。 原文: Detecting an intertextual reference is only the first step; philologists also need to know how a later author reworked the source. We explain a reuse pair word by word: each word of the reusing passage is first aligned to its source word, then labeled with the operation that transformed it, such as a copy, a change of inflection, or a substitution. Such word-level operations are expensive to annotate, and only a few corpora annotated at this level exist for text reuse in Latin literature. We describe a procedure that needs only a limited number of annotations: synthetic pairs, built by applying known operations to Latin text, train a first model, and active learning chooses the pairs an annotator corrects. Among several architectures, a model that links and labels together performs best: a pointer network that, for every reuse word, chooses a source word and names the operation in the same step. We show that with active learning, 200 pairs chosen by uncertainty are enough for the model to match one trained on every annotated pair of the 1,490 references of the Loci Similes benchmark. We release our code, word-level predictions for all references, and an interactive demo. 拆分:

    词级对齐与标注

    Detecting an intertextual reference is only the first step; philologists also need to know how a later author reworked the source. We explain a reuse pair word by word: each word of the reusing passage is first aligned to its source word, then labeled with the operation that transformed it, such as a copy, a change of inflection, or a substitution.

    标注挑战与流程设计

    Such word-level operations are expensive to annotate, and only a few corpora annotated at this level exist for text reuse in Latin literature. We describe a procedure that needs only a limited number of annotations: synthetic pairs, built by applying known operations to Latin text, train a first model, and active learning chooses the pairs an annotator corrects.

    模型架构与实验结果

    Among several architectures, a model that links and labels together performs best: a pointer network that, for every reuse word, chooses a source word and names the operation in the same step. We show that with active learning, 200 pairs chosen by uncertainty are enough for the model to match one trained on every annotated pair of the 1,490 references of the Loci Similes benchmark.

    资源发布

    We release our code, word-level predictions for all references, and an interactive demo.

    检查是否增删改: "Detecting an intertextual reference is only the first step; philologists also need to know how a later author reworked the source. We explain a reuse pair word by word