Keyword Selection Methodology for Identification of Major Events using Social Networks

The understanding of information communicated over social networks enables quick tracking of real events as they occur. In other cases, where the “crowd” factor is on high note, it is possible to identify events and to evaluate their magnitude, even before they occur. A full assessment of the content generated by social network users is very complex. This, due to the gigantic volume of data communicated over the net at any given time. Using few, well defined, keywords for the detection of relevant data reduces, considerably, the processing effort and expedites the identification of events, such as wildfire, floods or terror attacks. The preliminary results here has shown that by using keywords, specially tailored for different types of major events, one may detect ‘abnormal' surges of social network activities. Also, presented are threshold values, in terms of magnitude and frequency designed for early detection of these events. This approach is the basis for the development of algorithms for early identification real time systems and for geographical tracking of major events.