Scholarlink

<p>Automatic keyword extraction from academic papers is a pivotal task in natural language processing and information retrieval. While previous research has&#xa0;primarily focused on utilizing abstract, title, or references, this paper focuses on the highlights section—a structured&#xa0;summary delineating key findings and contributions&#xa0;to provide readers with&#xa0;a rapid overview of the research. Our observations indicate that highlights contain valuable keyword information that can effectively complement the abstract. To investigate the impact of incorporating highlights into unsupervised keyword extraction, we compared three input scenarios: using only the abstract, the highlights, and a combination of both. Experiments conducted using four unsupervised models on Computer science (CS), Library and Information Science (LIS) datasets reveal that incorporating the abstract with highlights significantly improves extraction performance. Furthermore, we analyze the differences in keyword coverage and content between abstract and highlights&#xa0;to elucidate how these variations influence extraction outcomes. The data and code are available at <a href="https://github.com/xiangyi-njust/Highlight-KPE">https://github.com/xiangyi-njust/Highlight-KPE</a>.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing unsupervised keyword extraction in academic papers through integrating highlights with abstract

  • 期刊论文

  • 发表日期：

    2026年4月30日

查看全文
  • Yi Xiang,
  • Chengzhi Zhang

摘要

Automatic keyword extraction from academic papers is a pivotal task in natural language processing and information retrieval. While previous research has primarily focused on utilizing abstract, title, or references, this paper focuses on the highlights section—a structured summary delineating key findings and contributions to provide readers with a rapid overview of the research. Our observations indicate that highlights contain valuable keyword information that can effectively complement the abstract. To investigate the impact of incorporating highlights into unsupervised keyword extraction, we compared three input scenarios: using only the abstract, the highlights, and a combination of both. Experiments conducted using four unsupervised models on Computer science (CS), Library and Information Science (LIS) datasets reveal that incorporating the abstract with highlights significantly improves extraction performance. Furthermore, we analyze the differences in keyword coverage and content between abstract and highlights to elucidate how these variations influence extraction outcomes. The data and code are available at https://github.com/xiangyi-njust/Highlight-KPE.