[论文解读] Public Sentiment and Demand for Used Cars after A Large-Scale Disaster: Social Media Sentiment Analysis with Facebook Pages

Yuya Shibuya, Hideyuki Tanaka|arXiv (Cornell University)|Jan 22, 2018

Public Relations and Crisis Communication参考文献 13被引用 4

一句话总结

本研究分析了2011年日本东北大地震及海啸后，Facebook页面上的公众情绪，并将社交媒体情绪与实际的二手车价格联系起来。研究发现，基于本地地理信息的沟通所产生的情绪与二手车价格正相关，而基于非本地灾难兴趣的沟通所产生的情绪则与二手车价格负相关，表明在长期恢复过程中，线上情绪对现实世界经济行为的影响因来源不同而存在差异。

ABSTRACT

There have been various studies analyzing public sentiment after a large-scale disaster. However, few studies have focused on the relationship between public sentiment on social media and its results on people's activities in the real world. In this paper, we conduct a long-term sentiment analysis after the Great East Japan Earthquake and Tsunami of 2011 using Facebook Pages with the aim of investigating the correlation between public sentiment and people's actual needs in areas damaged by water disasters. In addition, we try to analyze whether different types of disaster-related communication created different kinds of relationships on people's activities in the physical world. Our analysis reveals that sentiment of geo-info-related communication, which might be affected by sentiment inside a damaged area, had a positive correlation with the prices of used cars in the damaged area. On the other hand, the sentiment of disaster-interest-based-communication, which might be affected more by people who were interested in the disaster, but were outside the damaged area, had a negative correlation with the prices of used cars. The result could be interpreted to mean that when people begin to recover, used-car prices rise because they become more positive in their sentiment. This study suggests that, for long-term disaster-recovery analysis, we need to consider the different characteristics of online communication posted by locals directly affected by the disaster and non-locals not directly affected by the disaster.

研究动机与目标

• 调查大规模灾难发生后，社交媒体情绪与现实世界经济行为之间的关系。
• 研究不同类型的灾难相关在线沟通如何影响现实世界的结果，如二手车价格。
• 区分由直接受灾本地居民产生的情绪与仅具外围兴趣的非本地人群产生的情绪。
• 通过识别线上情绪如何反映或影响实际经济需求，为长期灾难恢复分析做出贡献。

提出的方法

• 对与2011年日本东北大地震及海啸相关的Facebook页面进行了长期情绪分析。
• 将社交媒体内容分类为两类：基于地理信息的沟通（本地、受灾相关）与基于灾难兴趣的沟通（非本地、外部驱动）。
• 使用自然语言处理技术，随时间收集并分析Facebook页面的情绪得分。
• 将情绪趋势与受灾地区实际的二手车价格进行相关性分析，以识别时间上的关联。
• 使用时间序列分析，比较情绪动态与经济指标的关系，重点关注灾难后的恢复阶段。

实验结果

研究问题

• RQ1 Facebook页面上表达的情绪与受灾地区二手车价格变化之间有何关系？
• RQ2 本地居民的情绪模式与非本地观察者的情绪模式相比，对现实世界经济活动的影响是否不同？
• RQ3 大规模灾难发生后，线上情绪变化与二手车需求变化之间的时间关系是什么？
• RQ4 不同类型的灾难相关沟通所产生的情绪能否预测现实世界经济行为的变化？

主要发现

• 反映本地实地经历的基于地理信息的沟通所产生的情绪，与受灾地区的二手车价格呈正相关。
• 由外部非受灾人群驱动的基于灾难兴趣的沟通所产生的情绪，与二手车价格呈负相关。
• 正相关表明，随着恢复过程中本地情绪改善，二手车需求上升，反映出经济活动的复苏。
• 负相关表明，非本地兴趣（常由媒体或好奇心驱动）可能与本地经济信心下降或投机行为相关。
• 结果表明，线上情绪并非单一整体；其对现实世界行为的影响，关键取决于情绪的来源与语境。