Abstract

In this article we present the foundations of a decision support system for blockage management in Fire Service. Blockage refers to the situation when all fire units are out and a new incident occurs. The approach is based on two phases: off-line data preparation and online blockage estimation. The off-line phase consists of methods from data mining and natural language processing and results in semantically coherent information granules. The online phase is about building the probabilistic models that estimate the block-age probability based on these granules. Finally, the selected classifier judges whether a blockage can occur and whether the resources from neighbour fire stations should be asked for assistance.