Sample of 376 article texts

The dataset comprises a collection of scientific articles, each represented by its full text and abstract, alongside the number of sentences in the abstract. The focus of the research utilizing this dataset is on optimizing the text summarization process, specifically honing in on the 'min_df' parameter, which is crucial for filtering terms in the summarization algorithm. Although the dataset contains various other fields, the analysis primarily utilized the article texts, abstract texts, and the count of sentences in the abstracts. This streamlined approach is aimed at enhancing the extractive summarization's effectiveness, judged by the ROUGE-1 score, a common metric for evaluating the quality of summarized texts. The objective is to fine-tune the summarization tool to produce high-quality summaries that are both informative and reflective of the original text, thereby improving the tool's utility in processing scientific documents.

本数据集囊括了一系列科学文献，每篇文献均以其全文及摘要为表征，并附带摘要中的句子数量。研究团队利用本数据集的主要目标是优化文本摘要过程，特别是对摘要算法中至关重要的 'min_df' 参数进行精炼。尽管数据集包含众多其他字段，但分析主要集中于文章全文、摘要文本以及摘要中的句子计数。这种简化的方法旨在提升抽提式摘要的有效性，以ROUGE-1分数作为衡量标准，ROUGE-1是评估摘要文本质量的一种常用指标。研究目标是对摘要工具进行微调，以生成既具有信息性又忠实于原文的高质量摘要，从而增强该工具在处理科学文献方面的实用性。