Urban Crisis Detection Technique: A Spatial and Data Driven Approach Based on Latent Dirichlet Allocation (LDA) Topic Modeling
Urban Crisis Detection Technique: A Spatial and Data Driven Approach Based on Latent Dirichlet Allocation (LDA) Topic Modeling
复制标题
城市危机检测技术：基于潜在狄利克雷分配（LDA）主题建模的空间和数据驱动方法
DOI：
10.1061/9780784481271.025
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
Taylor, John E.
中科院分区：
文献类型：
--
作者：
Wang, Yan;Taylor, John E.
社交网络平台已被广泛用于检测、跟踪和可视化人口密集的城市地区的物理事件。它们可以成为追溯或实时了解何时、何地以及发生了什么的有效工具。相应地，已经提出了多种方法来检测目标事件或一般事件。然而，这两种类型的事件检测技术都尚未开发用于检测特定地理位置和具有不可预测特征的城市灾害。因此，我们提出了一种空间和数据驱动的技术来检测城市灾害。该方法解决了事件的地理和语义维度（地理主题检测模块），并根据负面情绪的强度评估其危机级别（排名模块）。我们的方法是专门为地理参考推文设计的。为了演示该系统，我们在伦敦对带有地理标记的推文进行了 4 小时的实验。我们的城市危机检测技术成功地在所有候选地理主题中识别出了格伦菲尔大厦火灾。我们未来的工作重点是在大量流数据中实现具有高可扩展性的在线模式检测。已完成的研究将有助于在危机检测和跟踪、态势感知和信息传播方面提高灾害信息学和城市复原力。
Social networking platforms have been widely employed to detect, track, and visualize physical events in population-dense urban areas. They can be effective tools to understand when, where, and what happens retrospectively or in real time. Correspondingly, a variety of approaches have been proposed for detecting either targeted or general events. However, neither type of event detection technique has been developed to detect urban disasters in specific geographic locations and with unpredictable characteristics. Therefore, we propose a spatial and data-driven technique for detecting urban disasters. The method addresses both geographical and semantical dimensions of events (geo-topic detection module) and evaluates their crisis levels based on the intensity of negative sentiment (ranking module). Our approach was designed specifically for georeferenced tweets. To demonstrate the system, we conducted an experiment with 4-h of geotagged tweets in London. Our urban crisis detection technique successfully identified the Grenfell Tower fire among all the candidate geo-topics. Our future work focuses on enabling online-mode detection with high scalability in large-volumes of streaming data. The completed research will contribute to efficient disaster informatics and urban resilience regarding crisis detection and tracking, situation awareness, and information diffusion.