Spatial Cognition and Local Knowledge in Open Space: Ontologies in risk situations

Abstract

Recently, the diffusion of Social Networks is opening new research scenarios in risk assessment. In an emergency, during critical events, massive flows of information (text messages) posted on Social Networks could contribute to save lives or to help people in danger – provided they were tapped into and correctly interpreted by emergency agencies. These potential sources of information, in most cases, consist in unstructured social contents reflecting people’s intentions, perceptions and needs and they often have elements of complexity and uncertainty, hindering interpretation and thus thwarting response management.
The text messages are in natural language; they frequently contain locational information which, if properly extracted and processed, could make a key contribution to disaster management, and search and rescue in particular.
This research aims to contribute to understanding, in the context of social streaming analysis in a risk situation, how locational information and other implicit spatial knowledge may be organized to be effectively shared between all actors involved in disaster management. To that aim, an integrated approach involving machine learning and ontological models has been tested to help discover spatial knowledge.