Scholarlink

<p>Urban commuters in India often face transportation challenges during their daily travels. Traditional feedback methods, such as surveys and hotlines, struggle to scale effectively due to the large population in Indian cities. In this context, social media platforms such as Twitter/X present a practical alternative. where commuters’ complaints are often voiced through short, informal posts. These commuter complaints represent various ongoing issues in India’s urban transportation sector. Hence they are important for urban planners, policymakers, and transportation authorities to gain real-time insights into public concerns. However, an efficient framework is needed to automatically identify transportation-related concerns from the vast pool of social media posts and then generate a concise summary highlighting the most pressing issues, so that the policymakers/authorities can understand the key challenges and respond effectively to them. This study proposes a framework that utilizes generative AI and Natural Language Processing (NLP) to automatically identify and summarize transportation-related complaints from social media posts. To improve the quality of summarization, a novel prompt is developed for systematically summarizing transportation-related concerns and grievances. Findings indicate that this prompt significantly enhances summarization performance with the GPT-4 Turbo LLM. Notably, GPT-4-Turbo using proposed prompt achieves a ROUGE score of 0.86, surpassing the widely used LexRank algorithm, which scores 0.45.</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

Hear the Commute: A Generative AI-Based Framework to Summarize Transport Grievances from Social Media

  • 期刊论文

  • 发表日期：

    2025年5月13日

查看全文
  • Rahul Pullanikkat,
  • Moumita Basu,
  • Saptarshi Ghosh

摘要

Urban commuters in India often face transportation challenges during their daily travels. Traditional feedback methods, such as surveys and hotlines, struggle to scale effectively due to the large population in Indian cities. In this context, social media platforms such as Twitter/X present a practical alternative. where commuters’ complaints are often voiced through short, informal posts. These commuter complaints represent various ongoing issues in India’s urban transportation sector. Hence they are important for urban planners, policymakers, and transportation authorities to gain real-time insights into public concerns. However, an efficient framework is needed to automatically identify transportation-related concerns from the vast pool of social media posts and then generate a concise summary highlighting the most pressing issues, so that the policymakers/authorities can understand the key challenges and respond effectively to them. This study proposes a framework that utilizes generative AI and Natural Language Processing (NLP) to automatically identify and summarize transportation-related complaints from social media posts. To improve the quality of summarization, a novel prompt is developed for systematically summarizing transportation-related concerns and grievances. Findings indicate that this prompt significantly enhances summarization performance with the GPT-4 Turbo LLM. Notably, GPT-4-Turbo using proposed prompt achieves a ROUGE score of 0.86, surpassing the widely used LexRank algorithm, which scores 0.45.