Web data mining of landslide information, an experimental study for Macedonia

The current paper introduces an innovative method to generate a landslide database utilising the possibilities offered by web data mining techniques. As sources of data are used social media networks and news aggregators, in combination with already available landslide databases of the authors. The algorithm of the data mining is explained, as well as the rules for relevant classification and recognition of landslide information. The territory of Macedonia is taken as a case study to test the method. Findings show that the approach can be very useful from various aspects, with main advantage being the swift landslide data collection and possibility for sharing among relevant institutions. The approach is also considered promising in regard to timely alarming of the population for expected danger from landslides. Certain aspects of possibilities for mining additional web content related to landslides are noted, as well as options for integration of the web mined landslide data with already existing databases and sharing among entities. Due to the advantage of language recognition during the data mining, the utilisation of the presented method is possible on both national and international level.