Scholarlink

In this research paper, we delve into different methods of text summarization using LangChain and OpenAI LLM models. While performing summarization, it is hard for the user to find out which method will provide the best summary for their particular text. To solve this problem, five different methods of text summarization, namely basic prompt summarization, summarization using prompt template, stuff documents chain, map reduce method and refine technique are discussed in detail and implemented. Text summarization is performed on various speeches by great political leaders of India using these methods. Each model’s performance is evaluated and compared using BERT Score. For summarizing short texts, prompt template summarization performs best. While for summarizing lengthier multi-page documents, refine technique proves best.

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

Comparative Study of Methods for Text Summarization Using Generative AI

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Riza Atik,
  • Ritu Rani,
  • Garima Jaiswal,
  • Rajiv Sharma,
  • Arun Sharma

摘要

In this research paper, we delve into different methods of text summarization using LangChain and OpenAI LLM models. While performing summarization, it is hard for the user to find out which method will provide the best summary for their particular text. To solve this problem, five different methods of text summarization, namely basic prompt summarization, summarization using prompt template, stuff documents chain, map reduce method and refine technique are discussed in detail and implemented. Text summarization is performed on various speeches by great political leaders of India using these methods. Each model’s performance is evaluated and compared using BERT Score. For summarizing short texts, prompt template summarization performs best. While for summarizing lengthier multi-page documents, refine technique proves best.