Title: Gathering High Quality Information on Landslides from Twitter by Relevance Ranking of Users and Tweets
Abstract: Social networking platforms are increasingly used to report or pass along news and other valuable information. Their use rises especially during emergency situations and can be monitored for the analysis of adverse events, such as disasters. In this paper, we provide an overview of a comprehensive disaster information system using social networks with landslides serving as an illustrative example. We briefly describe each of the steps involved and focus on the classification and ranking steps that determine the relevance of individual messages and groups of messages to landslides. We introduce the concept of "relevant" and "irrelevant" virtual communities of users and compute their influence in each of them. This allows us to improve the existing relevance ranking formula by taking into account not only the semantics of the messages posted by users, but also the users' influence and the amount of their activity in these communities to improve the quality of the collected information on landslides. The resulting system achieves 0.936 F1-score in classifying individual messages and 0.941 F1-score in relevance ranking of the events.
Publication Year: 2016
Publication Date: 2016-11-01
Language: en
Type: article
Indexed In: ['crossref']
Access and Citation
Cited By Count: 13
AI Researcher Chatbot
Get quick answers to your questions about the article from our AI researcher chatbot