Sample of 376 article texts

Published: 19 March 2024| Version 1 | DOI: 10.17632/6zx6fw5t4t.1
Shakarim Aubakirov


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.



Kazakh British Technical University


Text Extraction