Giving meaning to tweets in emergency situations: a semantic approach for filtering and visualizing social data

In this paper, we propose a semantic approach for monitoring information published on social networks about a specific event. In the era of Big Data, when an emergency occurs information posted on social networks becomes more and more helpful for emergency operators. As direct witnesses of the situation, people share photos, videos or text messages about events that call their attention. In the emergency operation center, these data can be collected and integrated within the management process to improve the overall understanding of the situation and in particular of the citizen reactions. To support the tracking and analyzing of social network activities, there are already monitoring tools that combine visualization techniques with geographical maps. However, tweets are written from the perspective of citizens and the information they provide might be inaccurate, irrelevant or false. Our approach tries to deal with data relevance proposing an innovative ontology-based method for filtering tweets and extracting meaningful topics depending on their semantic content. In this way data become relevant for the operators to make decisions. Two real cases used to test its applicability showed that different visualization techniques might be needed to support situation awareness. This ontology-based approach can be generalized for analyzing the information flow about other domains of application changing the underlying knowledge base.