Discourse processing suffers from data sparsity, especially for dialogues. As a result, we explore approaches to build discourse structures for dialogues, based on attention matrices from Pre-trained Language Models (PLMs). We investigate multiple tasks for fine-tuning and show that the dialogue-tailored Sentence Ordering task performs best. To locate and exploit discourse information in PLMs, we propose an unsupervised and a semi-supervised method. Our proposals achieve encouraging results on the STAC corpus, with F1 scores of 57.2 and 59.3 for unsupervised and semi-supervised methods, respectively. When restricted to projective trees, our scores improved to 63.3 and 68.1.

翻译：语篇处理面临数据稀疏问题，尤其在对话场景中。为此，我们探索基于预训练语言模型（PLMs）的注意力矩阵构建对话语篇结构的方法。通过研究多种微调任务，我们发现针对对话优化的句子排序任务表现最佳。为定位并利用PLMs中的语篇信息，我们提出无监督与半监督两种方法。在STAC语料库上的实验取得令人鼓舞的结果：无监督方法和半监督方法的F1分数分别达到57.2和59.3；当限定为投影树时，分数提升至63.3和68.1。