The increasing capability and widespread usage of large language models (LLMs) highlight the desirability of automatic detection of LLM-generated text. Zero-shot detectors, due to their training-free nature, have received considerable attention and notable success. In this paper, we identify a new feature, token cohesiveness, that is useful for zero-shot detection, and we demonstrate that LLM-generated text tends to exhibit higher token cohesiveness than human-written text. Based on this observation, we devise TOCSIN, a generic dual-channel detection paradigm that uses token cohesiveness as a plug-and-play module to improve existing zero-shot detectors. To calculate token cohesiveness, TOCSIN only requires a few rounds of random token deletion and semantic difference measurement, making it particularly suitable for a practical black-box setting where the source model used for generation is not accessible. Extensive experiments with four state-of-the-art base detectors on various datasets, source models, and evaluation settings demonstrate the effectiveness and generality of the proposed approach. Code available at: \url{https://github.com/Shixuan-Ma/TOCSIN}.

翻译：大语言模型（LLM）能力的日益增强及其广泛使用，突显了对LLM生成文本进行自动检测的必要性。零样本检测器因其无需训练的特性，已受到广泛关注并取得了显著成功。本文提出一种适用于零样本检测的新特征——词元内聚性，并证明LLM生成文本往往比人类书写文本表现出更高的词元内聚性。基于这一发现，我们设计了TOCSIN——一种通用的双通道检测范式，该范式将词元内聚性作为即插即用模块来改进现有的零样本检测器。为计算词元内聚性，TOCSIN仅需进行数轮随机词元删除及语义差异度量，这使其特别适用于生成源模型不可访问的实际黑盒场景。通过在多样化数据集、源模型及评估设置下对四种前沿基础检测器进行大量实验，验证了所提方法的有效性与普适性。代码发布于：\url{https://github.com/Shixuan-Ma/TOCSIN}。