VietnameseMDS

该数据集包含了200个聚类，每个聚类代表从Baomoi收集的越南新闻话题，通常每个聚类包含2到5篇相关文档。数据集不仅包括句子和单词级别的分词数据，还提供了两份由人类生成的黄金参考摘要，用于评估。此外，该数据集中的平均句子长度为50个单词，并用于与人工生成的摘要进行ROUGE评分比较，以衡量提取式文本摘要任务的性能。规模上，数据集包含200个聚类，每个聚类有2至5篇文档，适用于提取式文本摘要任务。

This dataset comprises 200 clusters, each representing a Vietnamese news topic collected from Baomoi, with typically 2 to 5 relevant documents per cluster. The dataset not only includes sentence-level and word-level tokenized data, but also provides two human-generated gold standard reference summaries for evaluation purposes. Furthermore, the average sentence length of this dataset is 50 words, which is utilized to compare with human-written summaries via ROUGE scoring to assess the performance of extractive text summarization tasks. In terms of scale, the dataset contains 200 clusters with 2 to 5 documents per cluster, and is tailored for extractive text summarization tasks.