LexLattice: Multilingual Extractive Summarization via Neural Cellular Automata on Document Hierarchies

LexLattice：基于文档层次结构中神经元胞自动机的多语言抽取式摘要

Abstract: Faithfulness is a central concern in legal text summarization, which motivates extractive approaches that select verbatim content traceable to its source. Such methods typically rank paragraphs or other structural units in isolation, yet give little attention to consolidating evidence that is distributed across, and shares salience between, distant parts of a document.

摘要： 忠实度是法律文本摘要中的核心关注点，这促使人们采用抽取式方法，即选择可追溯至源头的逐字内容。此类方法通常孤立地对段落或其他结构单元进行排序，却很少关注如何整合分布在文档不同部分且具有共同显著性的证据。

We introduce LexLattice, an extractive summarizer that reifies a legal act’s hierarchy as a two-dimensional semantic lattice and consolidates over it with a masked 2D neural cellular automata before selection.

我们引入了 LexLattice，这是一种抽取式摘要器，它将法律法案的层次结构具体化为一个二维语义网格，并在选择前利用掩码二维神经元胞自动机（Neural Cellular Automata）对其进行整合。

LexLattice attains state-of-the-art ROUGE across all 24 languages of EUR-Lex-Sum in both multilingual and cross-lingual settings, surpassing instruction-tuned baselines with billions of parameters, despite concentrating all trainable capacity in a 1.8M parameter consolidator over a frozen multilingual encoder.

LexLattice 在多语言和跨语言设置下，在 EUR-Lex-Sum 的全部 24 种语言中均达到了最先进的 ROUGE 指标。尽管其所有可训练参数仅集中在一个 180 万参数的整合器中，并基于冻结的多语言编码器运行，但它仍超越了拥有数十亿参数的指令微调基线模型。

A consolidator trained only on high-resource languages further transfers to unseen languages with near-lossless retention (0.99), indicating that the model operates on language-agnostic semantic geometry rather than surface form.

仅在资源丰富语言上训练的整合器，在迁移到未见过的语言时表现出近乎无损的保留率（0.99），这表明该模型是基于与语言无关的语义几何结构，而非表面形式进行运作的。

Our results position explicit consolidation over document structure as a compact and traceable alternative to scale for multilingual legal summarization.

我们的研究结果表明，针对文档结构进行显式整合，是实现多语言法律摘要规模化的一种紧凑且可追溯的替代方案。