Publications / CCC 2025 - Zadar, Croatia

EXPLORING SOCIAL CONFLICT SPREAD PATTERNS FOR PUBLIC RAILWAY PROJECTS USING LARGE LANGUGE MODEL

Jawon Hong, Seungwon Baek, Sanghyun Lee, Seung Heon Han
Pages 384-392 (CCC 2025 - Zadar, Croatia, ISBN 978-1-7643710-0-1, ISSN 2413-5844)
Abstract:

Social conflicts in public construction projects have become increasingly complex due to growing involvement of diverse stakeholder and broader societal attention. These conflicts can escalate from regional and spread to national level unless they are unmanaged and can be significant risk to project success. With advancements in Natural Language Processing, previously hard-to-obtain data can now be efficiently processed, offering new opportunities for conflict analysis in the construction field. This study aims to identify social conflict diffusion patterns in public construction through news article analysis. Two similar public urban railway projects in South Korea were selected for comparative case analysis. The authors employ 3-phase news article data processing: (1) web crawling of regional and national news articles, (2) duplicate removal and keyword filtering, and (3) classification using GPT-4o model based on 16 conflict drivers identified from literature review. The findings revealed the importance of early conflict detection and proactive governance in mitigating social conflicts in infrastructure projects. This study demonstrates the potential usage of NLP techniques in systematically analysing large-scale textual data to identify patterns in conflict emergence, escalation and decrement. By identifying key conflict drivers and their diffusion patterns, the findings provide valuable insights for proactive conflict management, emphasizing the importance of early intervention and transparent communication to prevent regional issues from spread into broader societal challenges.

Keywords: conflict management, natural language processing, social conflict diffusion, public construction project