/
划分小节，那原文里有没有需要划分的地方呢？原文一开始讲分子子结构的重要性，然后提到现有方法的局限性，接着说提出的方法，再讲多尺度表示和模型构建，最后是实验和解释。可能可以把“然而现有方法的局限性”和“提出的方法”作为小节标题，用
来划分，因为这是两个主要的逻辑转折部分。然后每个自然段落用
包裹，原文里的列表内容？看起来原文里没有明显的列表，不过引用的话？原文里没有明显的引用句子，但重要的是要保留所有信息，不能增删改。然后重要句子用，比如原文里的“propose a geometry-based fragmentation method that partitions arbitrary molecular or material structures using largely geometric information, without relying on predefined chemical fragmentation rules”这个句子是核心方法，应该用包裹。还有“Experiments on a benchmark polymer dataset show that incorporating 3D conformational information improves prediction accuracy for several properties over models based exclusively on p-SMILES representations”这个也是重要结论，也可以用。接下来，处理段落划分：原文第一部分是从开头到“Fragment-based representations have therefore emerged as an effective strategy for incorporating chemically meaningful inductive biases, as exemplified by molecular fingerprints and fragment-based attention mechanisms.”这是一个段落，用
包裹。然后下一段落是“However, many existing fragmentation-based machine learning methods rely on predefined chemical rules, which may yield suboptimal fragmentation for previously unseen or chemically unfamiliar structures and often overlook the importance of multiscale molecular representations.”这个可以用
作为小节标题，标题内容就是“Limitations of existing methods”之类的？不对，规则说不能增删改内容，所以小节标题必须从原文里提取。原文里“However...”这段是现有方法的问题，所以标题应该用“Limitations of existing fragmentation-based methods”？但原文里没有这个标题，规则4说不得增删改内容，所以不能添加原文里没有的标题文字？哦，等一下，规则2说“长文用
/
划分小节”，但小节标题必须是原文里的内容吗？比如原文里“To address these limitations, we formulate molecular fragmentation...”这部分是提出的方法，那小节标题可以用“Proposed formulation and method”？但原文里没有这个标题，那是不是可以不用
，或者把原文里的长句拆分成小节标题？比如原文里“However...”后面的句子是讲现有方法的问题，可以把这个作为小节标题，但必须是原文里的原句？不对，规则4说不得增删改内容，所以不能添加任何原文没有的文字作为标题。那可能原文的结构里没有明显的小节标题，那是不是可以不用
和
？但规则2说“长文用
/
划分小节”，如果原文是长文的话。原文大概有几百字，是长文，所以需要划分小节。那怎么从原文里提取小节标题呢？比如原文里的“To address these limitations, we formulate molecular fragmentation as a radius-constrained optimization problem...”这部分是方法介绍，可以把这部分之前的内容作为第一个小节，之后的是第二个小节。比如第一个小节标题是原文里的“It is well established that molecular substructures can contribute unequally to chemical and physical properties, with specific motifs or side chains often playing dominant roles. Fragment-based representations have therefore emerged as an effective strategy for incorporating chemically meaningful inductive biases, as exemplified by molecular fingerprints and fragment-based attention mechanisms.”但这是一个段落，不能作为标题。或者是不是我理解错了，规则2说“长文用
/
划分小节”，可能原文里的内容可以拆分成几个部分，每个部分用
或
作为小节标题，标题内容必须是原文里的原句或原词？比如原文里“However, many existing fragmentation-based machine learning methods rely on predefined chemical rules, which may yield suboptimal fragmentation for previously unseen or chemically unfamiliar structures and often overlook the importance of multiscale molecular representations.”这句话可以作为小节标题吗？但这句话是一个句子，作为标题太长了。或者，可能原文里的内容不需要强行划分小节，因为没有明显的小节标题，所以只用
包裹每个自然段落。那原文里的自然段落有几个？原文看起来是一个连续的段落，但其实可以拆分成几个逻辑段落：第一个段落是前两句，讲分子子结构的重要性和片段表示的优势；第二个段落是“However...”到“scale-adaptive modeling.”讲现有方法的问题和提出的方法；第三个段落是“Building on this framework...”到最后，讲模型构建和实验结果。这样的话，每个逻辑段落用
包裹，然后重要的句子用。比如第一个段落：“It is well established that molecular substructures can contribute unequally to chemical and physical properties, with specific motifs or side chains often playing dominant roles. Fragment-based representations have therefore emerged as an effective strategy for incorporating chemically meaningful inductive biases, as exemplified by molecular fingerprints and fragment-based attention mechanisms.”这个用
包裹。第二个段落：“However, many existing fragmentation-based machine learning methods rely on predefined chemical rules, which may yield suboptimal fragmentation for previously unseen or chemically unfamiliar structures and often overlook the importance of multiscale molecular representations. To address these limitations, we formulate molecular fragmentation as a radius-constrained optimization problem inspired by the k-center formulation, where the number of clusters is not fixed a priori but emerges from the specified radius, and propose a geometry-based fragmentation method that partitions arbitrary molecular or material structures using largely geometric information, without relying on predefined chemical fragmentation rules. The method provides flexible control over fragment size, enabling the construction of multiscale fragment representations for scale-adaptive modeling.”这个用
包裹，里面的核心方法句子用。第三个段落：“Building on this framework, we develop a geometric fragment-based attention mechanism that integrates multiscale structural information and incorporate it into a multimodal machine learning model combining p-SMILES representations with three-dimensional (3D) conformational information. Experiments on a benchmark polymer dataset show that incorporating 3D conformational information improves prediction accuracy for several properties over models based exclusively on p-SMILES representations. Explainable artificial intelligence analyses further provide insights into the physical relevance of the learned representations.”这个用
包裹，里面的模型构建句子用。然后检查有没有遗漏的信息，所有句子都要保留。有没有引用？原文里没有明显的引用别人的话，所以不用
。有没有代码？没有，所以不用
。列表？没有，所以不用- 。小节标题？如果原文里没有明显的小节标题，是不是可以不用
/
？但规则2说“长文用
/
划分小节”，可能原文是长文，需要划分，那怎么划分呢？比如把“Limitations of existing methods”作为
，但这个标题是原文里没有的，违反规则4。所以可能原文里没有需要用
/
划分的部分，因为原文是一个连续的论述，没有明显的小节标题。所以最终的HTML片段应该是几个用
包裹的段落，里面的重要句子用包裹。比如：
It is well established that molecular substructures can contribute unequally to chemical and physical properties, with specific motifs or side chains often playing dominant roles. Fragment-based representations have therefore emerged as an effective strategy for incorporating chemically meaningful inductive biases, as exemplified by molecular fingerprints and fragment-based attention mechanisms.
However, many existing fragmentation-based machine learning methods rely on predefined chemical rules, which may yield suboptimal fragmentation for previously unseen or chemically unfamiliar structures and often overlook the importance of multiscale molecular representations. To address these limitations, we formulate molecular fragmentation as a radius-constrained optimization problem inspired by the k-center formulation, where the number of clusters is not fixed a priori but emerges from the specified radius, and propose a geometry-based fragmentation method that partitions arbitrary molecular or material structures using largely geometric information, without relying on predefined chemical fragmentation rules. The method provides flexible control over fragment size, enabling the construction of multiscale fragment representations for scale-adaptive modeling.
Building on this framework, we develop a geometric fragment-based attention mechanism that integrates multiscale structural information and incorporate it into a multimodal machine learning model combining p
。列表？没有，所以不用- 。小节标题？如果原文里没有明显的小节标题，是不是可以不用
/
？但规则2说“长文用
/
划分小节”，可能原文是长文，需要划分，那怎么划分呢？比如把“Limitations of existing methods”作为
，但这个标题是原文里没有的，违反规则4。所以可能原文里没有需要用
/
划分的部分，因为原文是一个连续的论述，没有明显的小节标题。所以最终的HTML片段应该是几个用
包裹的段落，里面的重要句子用包裹。比如：
It is well established that molecular substructures can contribute unequally to chemical and physical properties, with specific motifs or side chains often playing dominant roles. Fragment-based representations have therefore emerged as an effective strategy for incorporating chemically meaningful inductive biases, as exemplified by molecular fingerprints and fragment-based attention mechanisms.
However, many existing fragmentation-based machine learning methods rely on predefined chemical rules, which may yield suboptimal fragmentation for previously unseen or chemically unfamiliar structures and often overlook the importance of multiscale molecular representations. To address these limitations, we formulate molecular fragmentation as a radius-constrained optimization problem inspired by the k-center formulation, where the number of clusters is not fixed a priori but emerges from the specified radius, and propose a geometry-based fragmentation method that partitions arbitrary molecular or material structures using largely geometric information, without relying on predefined chemical fragmentation rules. The method provides flexible control over fragment size, enabling the construction of multiscale fragment representations for scale-adaptive modeling.
Building on this framework, we develop a geometric fragment-based attention mechanism that integrates multiscale structural information and incorporate it into a multimodal machine learning model combining p