Scholarlink

Text summarization is the process of distilling the most important information from a text document to produce a shorter version while retaining its key points and overall meaning. This paper presents detailed comparisons of different transformer-based models employed for long text summarization. Our study analyzes both abstractive summarization based models like LongT5, BigBird Pegasus, HAT-BART, DYLE and extractive summarization based models like BERTSUM, LED-Large, HETFORMER, HIStruct+, Long-Trand-Extr. By evaluating numerous studies, our assessment aims to reveal the effectiveness of these approaches and provide a nuanced understanding of their strengths and weaknesses. ROUGE1, ROUGE 2, and ROUGE L are employed as performance metrics in the analysis of respective models. This research serves as a comprehensive guide for professionals working with long document summarization, specifically focusing on various transformer-based approaches, paving the way for advancements in real-world applications.

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

A Comparative Analysis of Transformer-Based Models for Abstractive and Extractive Long Text Summarization

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Rubika Bashyal,
  • Bindu Paudel,
  • Nijiya Maharjan,
  • Sharad Kumar Ghimire

摘要

Text summarization is the process of distilling the most important information from a text document to produce a shorter version while retaining its key points and overall meaning. This paper presents detailed comparisons of different transformer-based models employed for long text summarization. Our study analyzes both abstractive summarization based models like LongT5, BigBird Pegasus, HAT-BART, DYLE and extractive summarization based models like BERTSUM, LED-Large, HETFORMER, HIStruct+, Long-Trand-Extr. By evaluating numerous studies, our assessment aims to reveal the effectiveness of these approaches and provide a nuanced understanding of their strengths and weaknesses. ROUGE1, ROUGE 2, and ROUGE L are employed as performance metrics in the analysis of respective models. This research serves as a comprehensive guide for professionals working with long document summarization, specifically focusing on various transformer-based approaches, paving the way for advancements in real-world applications.