Towards a Framework for the Automatic Detection of Crisis Emotions on Social Media: a Corpus Analysis of the Tweets Posted after the Crash of Germanwings Flight 9525

摘要

Social media, and in particular Twitter, are increasingly being utilized during crises. It has been shown that tweets offer valuable real-time information for decision-making. Given the vast amount of data available on the Web, there is a need for intelligent ways to select and retrieve the desired information. Analyzing sentiment and emotions in online text is one option for distinguishing relevant from irrelevant information. In this study, we investigate to what extent automatic sentiment analysis techniques can be used for detecting crisis emotions on Twitter. Therefore, a corpus of tweets posted after the crash of German-wings Flight 9525 was built and labeled with polarity and emotion information. Preliminary results show better classification results for the negative sentiment class compared to the positive class. An analysis of the more fine-grained emotion classification reveals that sympathy and anger are the most frequently expressed emotions in our corpus. To further enhance the performance of emotion classification in online crisis communication, it is crucial to accurately detect i) the object of the crisis emotion and ii) the characteristics of the sender.